DESIGN AND ANALYSIS OF A MEMS VIBRATION SENSOR FOR AUTOMOTIVE MECHANICAL
SYSTEMS
BY
Joel Rebello
A thesis submitted in conformity with the requirements
for the degree of Master of Applied Science
Graduate Department of Mechanical and Industrial Engineering
University of Toronto
copy Copyright by Joel Rebello (2009)
ABSTRACT
Design and Analysis of a MEMS Vibration Sensor for Automotive Mechanical Systems
Joel Rebello
Master of Applied Science
Graduate Department of Mechanical and Industrial Engineering
University of Toronto
2009
This thesis presents the theoretical analysis and experiment results of MEMS sensors
designed for the application of low frequency vibration sensing Each sensor consists of a
proof mass connected to a folded beam micro-flexure with an attached capacitive comb
drive for displacement sensing Three comb drive arrangements are evaluated the
transverse lateral and tri-plate differential The sensors are fabricated using the well
developed foundry processes of PolyMUMPS and SoiMUMPS In addition a capacitance
to voltage readout circuit is fabricated using discrete components Static tests evaluating
the capacitance to displacement relation are conducted on a six degree of freedom
robotic manipulator and dynamic tests evaluating the sensor response to sinusoidal
excitations are conducted on a vibrating beam The end use of the sensor involves real-
time vibration monitoring of automobile mechanical systems such as power seats
windshield wipers mirrors trunks and windows allowing for early detection of
mechanical faults before catastrophic failure
ii
ACKNOWLEDGMENT
I would first like to thank God for all his blessings and express my gratitude to my
parents and brother for their endless support love and encouragement
I would like to thank my supervisors Dr William Cleghorn and Dr James Mills for
their guidance I would also like to thank my friends and colleagues Henry Chu Dr
Pezhman Hassanpour Shael Markin Dr Lidai Wang and Dr Xuping Zhang at the
Laboratory for Nonlinear Systems Control for great conversation guidance and
friendship Thank you to the members of my Examination Committee Dr Foued Ben
Amara and Dr Jean Zu for their time and feedback
Finally I would like to extend my appreciation to CMC Microsystems for the MEMS
device fabrication and design support and to Autorsquo21 a Network of Federal Centers of
Excellence for project funding
iii
TABLE OF CONTENTS 1 INTRODUCTION 1
11 RESEARCH PURPOSE 1 12 MICROELECTROMECHANICAL SENSORS 2 13 AUTOMOTIVE SENSORS 3 14 OPERATING ENVIRONMENT 4 15 OBJECTIVES AND CONTRIBUTION 5 16 RESEARCH OUTLINE 5
2 LITERATURE REVIEW 7
21 EXISTING METHODS FOR FAULT DETECTION 7 211 Motor Current Monitoring 7 212 Temperature Sensors 8 213 Acoustic Emission Sensors 9
22 MEMS VIBRATION SENSORS 10 221 Resonant Sensors 10 222 Piezoelectric Sensors 11 223 Displacement Variation Sensors 12
3 SENSOR STRUCTURE 15
31 SENSING 15 32 MECHANICAL ANALYSIS 21
321 Micro-Flexure Selection 21 322 Equations of Motion 26 323 Quality Factor 30
33 DRIVE STABILITY 32 34 ELECTROSTATIC ACTUATION 34 35 SECTION SUMMARY 35
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT 36
5 NOISE ANALYSIS 40
51 MECHANICAL-THERMAL NOISE 40 52 ELECTRICAL NOISE 41
6 FABRICATION OVERVIEW 43
61 POLYMUMPS 43 62 SOIMUMPS 44
7 EXPERIMENTAL RESULTS 46
71 MEMS SENSOR ndash T1 46 72 READOUT CIRCUITRY 51 73 MEMS SENSORndash L1 52 74 SECTION SUMMARY 59
8 CONCLUSION AND FUTURE WORK 61
81 CONCLUSION 61 82 FUTURE WORK 62
iv
LIST OF TABLES TABLE 3-1 SPECIFICATIONS FOR EACH CAPACITIVE COMB DRIVE IMPLEMENTED IN THIS WORK T1 L1 AND
T2 21 TABLE 3-2 DIMENSIONS AND MECHANICAL PROPERTIES FOR THE THREE MEMS VIBRATION SENSORS
OUTLINED IN THIS WORK 30 TABLE 3-3 SUMMARY OF THE THEORETICAL COEFFICIENTS OF DAMPING AND QUALITY FACTORS FOR THE
THREE SENSORS PRESENTED IN THIS WORK 32 TABLE 3-4 SUMMARY OF THE CHARACTERISTICS OF THE SENSORS OUTLINED IN THIS WORK 35 TABLE 7-1 SUMMARY OF EXPERIMENTAL RESULTS 60
v
LIST OF FIGURES
FIGURE 3-1 AN INTER-DIGITATED COMB DRIVE WHERE CHANGES IN CAPACITANCE ARE GENERATED BY
EITHER CHANGES IN GAP DISTANCE X0 OR IN THE OVERLAP AREA Y0timesT 16 FIGURE 3-2 AN INTER-DIGITATED COMB DRIVE IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT WHERE X01
X02 AND X03 ARE THE GAP DISTANCES AND X01= X02ltlt X03 TO IMPROVE SENSITIVITY 18 FIGURE 3-3 COMPARISON OF THE EFFECT OF GAP DISTANCES ON LINEARITY FOR TWO CASE lsquorsquo REPRESENTS
CASE 1 WHERE THE FINGERS IN ARE IMPLEMENTED IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT AND X01= X02= 2 ΜM AND X03= 15 ΜM lsquorsquo REPRESENTS THE CASE 2 WHERE THE FINGERS ARE IMPLEMENTED IN A SINGLE ENDED ARRANGEMENT AND X01= X02=X03=2 ΜM 19
FIGURE 3-4 lsquorsquo REPRESENTS THE OUTPUT OF THE DIFFERENTIAL TRI-PLATE DRIVE WHILE lsquorsquo REPRESENTS THE OUTPUT OF THE TRANSVERSE DRIVE THE DASHED LINE REPRESENTS THE LINEAR APPROXIMATION FOR DISPLACEMENTS LESS THAT 1 ΜM FOR THE PURPOSE OF COMPARISON THE OUTPUT OF THE TRANSVERSE DRIVE WAS EQUALIZED TO ZERO AT ZERO DISPLACEMENT 20
FIGURE 3-5 SENSOR STRUCTURE 22 FIGURE 3-6 CLAMPED-CLAMPED FLEXURE 24 FIGURE 3-7 CRAB LEG FLEXURE 25 FIGURE 3-8 FOLDED BEAM FLEXURE 25 FIGURE 3-9 COMPARISON OF THE NON-LINEAR DEFLECTION AMONG THE THREE MICRO-FLEXURES
CONSIDERED CLAMPED-CLAMPED CRAB-LEG AND FOLDED BEAM ALL FLEXURES WERE DESIGNED TO HAVE THE SAME STIFFNESS OF 035 N M-1 26
FIGURE 3-10 LUMPED MASS APPROXIMATION WITH BASE EXCITATION YB(T) 27 FIGURE 3-11 LUMPED MASS APPROXIMATION WITH DIRECT MASS EXCITATION Z(T) 28 FIGURE 3-12 GRAPH REPRESENTING STABILITY REGIONS FOR GAP CLOSING CAPACITIVE DRIVES lsquorsquo
REPRESENTS THE RHS OF EQ (32) FOR T2 WHILE lsquorsquo REPRESENTS THE RHS OF T1 34 FIGURE 4-1 IDEAL CHARGE AMPLIFIER 36 FIGURE 4-2 CAPACITANCE TO VOLTAGE READOUT CIRCUIT WHICH IS MADE UP OF FOUR ELEMENTS THE
CHARGE AMPLIFIER VOLTAGE AMPLIFIER DEMODULATOR AND LOW PASS FILTER 37 FIGURE 4-3 ORCAD SIMULATION RESULTS OF THE CAPACITANCE TO VOLTAGE READOUT CIRCUIT THE
EXCITATION VOLTAGE VE IS 75 VPK AND THE OUTPUT OF THE VOLTAGE AMPLIFIER VVA IS APPROXIMATELY 13125 VPK 38
FIGURE 4-4 ORCAD SIMULATION RESULTS SHOWING THE SINUSOIDAL OUTPUT OF THE DEMODULATOR VDM WITH A 984 VPK-PK AMPLITUDE AND THE DC OUTPUT OF THE LOW PASS FILTER VLP AT APPROXIMATELY 5 V 39
FIGURE 5-1 ORCAD SIMULATION OF THE NOISE VOLTAGE SPECTRAL DENSITY ENO AT THE OUTPUT OF THE VOLTAGE AMPLIFIER 42
FIGURE 5-2 TOTAL OUTPUT NOISE ENO AT THE VOLTAGE AMPLIFIER WHICH IS EXPRESSED AS ENO=( int ENO2)12
42 FIGURE 6-1 POLYMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 44 FIGURE 6-2 SOIMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 45 FIGURE 7-1 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE VERSUS DISPLACEMENT
RELATION WHILE THE SOLID POINTS REPRESENT THE EXPERIMENTAL VALUES THE DASHED LINE REPRESENTS THE ASSUMED LINEAR RELATION FOR DISPLACEMENTS LESS THAN 1ΜM THE EXPERIMENTAL AVERAGE DRIVE SENSITIVITY IS APPROXIMATELY 0024 PF ΜM-1 FOR DISPLACEMENTS GREATER THAN 12 ΜM DRIVE INSTABILITY IS OBSERVED AND IS ATTRIBUTED TO THE PULL-IN EFFECT 47
FIGURE 7-2 lsquotimesrsquo REPRESENTS THE AMPLITUDE AND SOLID LINE REPRESENTS THE PHASE RESPONSE THE RESONANCE PEAK IS LOCATED AT 3600 HZ AND ACCOMPANIED BY A PHASE SHIFT OF 168deg 48
FIGURE 7-3 A SEM MICRO-GRAPH SHOWING THE SENSOR TI FABRICATED USING POLYMUMPS THE FOUR FOLDED BEAM FLEXURES HAD A LENGTH OF 650 ΜM AND TOTAL STIFFNESS OF 14 NM THE EFFECTIVE PROOF MASS WAS 29E-9 KG AND THE SYSTEM HAD A NATURAL FREQUENCY OF 3500 HZ 50
FIGURE 7-4 A SEM MICRO-GRAPH OF THE TRANSVERSE COMB DRIVE IMPLEMENTED IN T1 WITH 200 COMB FINGERS A GAP DISTANCE OF 2 ΜM OVERLAP LENGTH OF 45 ΜM AND THICKNESS OF 2 ΜM THE
vi
AVERAGE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 0024 PF ΜM-1 FOR DISPLACEMENTS OF LESS THAN 1 ΜM 50
FIGURE 7-5 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE TO VOLTAGE RELATION EXPRESSED BY EQ (37) WITH A SENSITIVITY OF 141 V PF-1 THE SOLID POINTS REPRESENT THE EXPERIMENTAL DATA POINTS WHEN DISCRETE CAPACITORS 1-4 PF ARE USED IN THE READOUT CIRCUIT THE HORIZONTAL DASHED LINE REPRESENTS THE OUTPUT VOLTAGE OF THE MEMS SENSOR AT REST AND CORRESPONDS TO A CAPACITANCE OF 35 PF 52
FIGURE 7-6 OSCILLOSCOPE TRACE OF THE INPUT EXCITATION VOLTAGE VE asymp 75 VPK AND OUTPUT OF THE VOLTAGE AMPLIFIER VVA asymp 103 VPK READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-7 OSCILLOSCOPE TRACE OF THE DEMODULATOR OUTPUT VDM asymp 76 VPK-PK AND OUTPUT OF THE LOW PASS FILTER VLP asymp 36 V READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-8 lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS WITH THE DASHED LINE REPRESENTING THE FITTED CURVE SOLID LINE REPRESENTS THE THEORETICAL DATA POINTS ACCORDING TO EQ (10) 54
FIGURE 7-9 EXPERIMENT SETUP WHICH INCLUDES THE MEMS SENOR AND READOUT BOARD MOUNTED ON THE CLAMPED-FREE BEAM AND TEST EQUIPMENT 55
FIGURE 7-10 CLOSE-UP OF THE MEMS SENSOR AND READOUT BOARD MOUNTED ON THE TEST BEAM ALONG WITH THE COMMERCIAL PIEZOELECTRIC ACCELEROMETER ONLY THE CHARGE AMPLIFIER (CA) AND THE VOLTAGE AMPLIFIER (VA) ARE MOUNTED ON THE SAME PROTOTYPE BOARD AS THE MEMS SENSOR DUE TO SPACE CONSTRAINTS 55
FIGURE 7-11 FIGURE COMPARING THE RESPONSE OF A COMMERCIAL PIEZOELECTRIC ACCELEROMETER AND MEMS SENSOR (L1) TO A BEAM VIBRATION A) DYNAMIC RESPONSE OF THE PIEZOELECTRIC ACCELEROMETER REPRESENTING THE TRUE BEAM VIBRATION WITH A PEAK VOLTAGE 114 V CORRESPONDING TO AN APPLIED ACCELERATION OF 114 G B) DYNAMIC RESPONSE OF THE MEMS SENSOR WITH A PEAK VOLTAGE OF 006 V 56
FIGURE 7-12 SENSOR DYNAMIC RESPONSE TO THE VIBRATION OF THE CLAMPED-FREE BEAM AND REPRESENTS A DIRECT EXPORT FROM THE DYNAMIC SIGNAL ANALYZER (AGILENT 35670) TOP TRACE IS THE TIME DOMAIN SIGNAL OF THE LOW PASS FILTER OUTPUT WHERE VLPMAXasymp73MV WHICH REPRESENTS A MAXIMUM ACCELERATION OF APPROXIMATELY 15G THE BOTTOM TRACE REPRESENTS THE FREQUENCY SPECTRUM WITH A PEAK AT 23HZ WHICH CORRESPONDS TO THE BEAM NATURAL FREQUENCY 57
FIGURE 7-13 EXPERIMENTAL RESULTS SHOWING THE RESPONSE OF THE SENSOR TO VARIOUS ACCELERATIONS IN THE RANGE OF 0-20 G WHERE lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS AND THE SOLID LINE REPRESENTS THE FITTED CURVE THE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 4258 MV G-1 58
vii
NOMENCLATURE a applied acceleration A applied acceleration amplitude App parallel plate overlapping area Asldie-film effective area for slide-film damping c coefficient of damping Cf feedback capacitance Cpp capacitance of the parallel plate capacitor cslide-film coefficient of damping as a result of slide-film damping
csqueeze-film coefficient of damping as a result of squeeze-film damping CTotal total capacitance of a comb drive CTotalBottom Total capacitance of the bottom drive in a tri-plate comb drive CTotalL total capacitance of a lateral drive CTotalT total capacitance of a transverse drive CTotalTop Total capacitance of the top drive in a tri-plate comb drive d distance between the parallel plates E Youngrsquos Modulus of sensor material eno noise voltage spectral density Eno total output noise voltage EnoAMP total output amplifier noise voltage EnoR total output resistor noise voltage
f excitation frequency in Hz fn system natural frequency in Hz
Fact electrostatic actuation force
FN noise force Fx-Elect total electrostatic force for pull-in instability Itruss second moment of inertia of truss section on the folded beam flexure Ibeam second moment of inertia of beam section on the folded beam
flexure kair dielectric constant of air at atmospheric pressure kB Boltzmannrsquos constant kT total stiffness of parallel folded beam flexure kx stiffness of various flexures in x-direction ky stiffness of various flexures in y-direction L flexure beam length Lplate effective plate length
viii
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
ABSTRACT
Design and Analysis of a MEMS Vibration Sensor for Automotive Mechanical Systems
Joel Rebello
Master of Applied Science
Graduate Department of Mechanical and Industrial Engineering
University of Toronto
2009
This thesis presents the theoretical analysis and experiment results of MEMS sensors
designed for the application of low frequency vibration sensing Each sensor consists of a
proof mass connected to a folded beam micro-flexure with an attached capacitive comb
drive for displacement sensing Three comb drive arrangements are evaluated the
transverse lateral and tri-plate differential The sensors are fabricated using the well
developed foundry processes of PolyMUMPS and SoiMUMPS In addition a capacitance
to voltage readout circuit is fabricated using discrete components Static tests evaluating
the capacitance to displacement relation are conducted on a six degree of freedom
robotic manipulator and dynamic tests evaluating the sensor response to sinusoidal
excitations are conducted on a vibrating beam The end use of the sensor involves real-
time vibration monitoring of automobile mechanical systems such as power seats
windshield wipers mirrors trunks and windows allowing for early detection of
mechanical faults before catastrophic failure
ii
ACKNOWLEDGMENT
I would first like to thank God for all his blessings and express my gratitude to my
parents and brother for their endless support love and encouragement
I would like to thank my supervisors Dr William Cleghorn and Dr James Mills for
their guidance I would also like to thank my friends and colleagues Henry Chu Dr
Pezhman Hassanpour Shael Markin Dr Lidai Wang and Dr Xuping Zhang at the
Laboratory for Nonlinear Systems Control for great conversation guidance and
friendship Thank you to the members of my Examination Committee Dr Foued Ben
Amara and Dr Jean Zu for their time and feedback
Finally I would like to extend my appreciation to CMC Microsystems for the MEMS
device fabrication and design support and to Autorsquo21 a Network of Federal Centers of
Excellence for project funding
iii
TABLE OF CONTENTS 1 INTRODUCTION 1
11 RESEARCH PURPOSE 1 12 MICROELECTROMECHANICAL SENSORS 2 13 AUTOMOTIVE SENSORS 3 14 OPERATING ENVIRONMENT 4 15 OBJECTIVES AND CONTRIBUTION 5 16 RESEARCH OUTLINE 5
2 LITERATURE REVIEW 7
21 EXISTING METHODS FOR FAULT DETECTION 7 211 Motor Current Monitoring 7 212 Temperature Sensors 8 213 Acoustic Emission Sensors 9
22 MEMS VIBRATION SENSORS 10 221 Resonant Sensors 10 222 Piezoelectric Sensors 11 223 Displacement Variation Sensors 12
3 SENSOR STRUCTURE 15
31 SENSING 15 32 MECHANICAL ANALYSIS 21
321 Micro-Flexure Selection 21 322 Equations of Motion 26 323 Quality Factor 30
33 DRIVE STABILITY 32 34 ELECTROSTATIC ACTUATION 34 35 SECTION SUMMARY 35
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT 36
5 NOISE ANALYSIS 40
51 MECHANICAL-THERMAL NOISE 40 52 ELECTRICAL NOISE 41
6 FABRICATION OVERVIEW 43
61 POLYMUMPS 43 62 SOIMUMPS 44
7 EXPERIMENTAL RESULTS 46
71 MEMS SENSOR ndash T1 46 72 READOUT CIRCUITRY 51 73 MEMS SENSORndash L1 52 74 SECTION SUMMARY 59
8 CONCLUSION AND FUTURE WORK 61
81 CONCLUSION 61 82 FUTURE WORK 62
iv
LIST OF TABLES TABLE 3-1 SPECIFICATIONS FOR EACH CAPACITIVE COMB DRIVE IMPLEMENTED IN THIS WORK T1 L1 AND
T2 21 TABLE 3-2 DIMENSIONS AND MECHANICAL PROPERTIES FOR THE THREE MEMS VIBRATION SENSORS
OUTLINED IN THIS WORK 30 TABLE 3-3 SUMMARY OF THE THEORETICAL COEFFICIENTS OF DAMPING AND QUALITY FACTORS FOR THE
THREE SENSORS PRESENTED IN THIS WORK 32 TABLE 3-4 SUMMARY OF THE CHARACTERISTICS OF THE SENSORS OUTLINED IN THIS WORK 35 TABLE 7-1 SUMMARY OF EXPERIMENTAL RESULTS 60
v
LIST OF FIGURES
FIGURE 3-1 AN INTER-DIGITATED COMB DRIVE WHERE CHANGES IN CAPACITANCE ARE GENERATED BY
EITHER CHANGES IN GAP DISTANCE X0 OR IN THE OVERLAP AREA Y0timesT 16 FIGURE 3-2 AN INTER-DIGITATED COMB DRIVE IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT WHERE X01
X02 AND X03 ARE THE GAP DISTANCES AND X01= X02ltlt X03 TO IMPROVE SENSITIVITY 18 FIGURE 3-3 COMPARISON OF THE EFFECT OF GAP DISTANCES ON LINEARITY FOR TWO CASE lsquorsquo REPRESENTS
CASE 1 WHERE THE FINGERS IN ARE IMPLEMENTED IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT AND X01= X02= 2 ΜM AND X03= 15 ΜM lsquorsquo REPRESENTS THE CASE 2 WHERE THE FINGERS ARE IMPLEMENTED IN A SINGLE ENDED ARRANGEMENT AND X01= X02=X03=2 ΜM 19
FIGURE 3-4 lsquorsquo REPRESENTS THE OUTPUT OF THE DIFFERENTIAL TRI-PLATE DRIVE WHILE lsquorsquo REPRESENTS THE OUTPUT OF THE TRANSVERSE DRIVE THE DASHED LINE REPRESENTS THE LINEAR APPROXIMATION FOR DISPLACEMENTS LESS THAT 1 ΜM FOR THE PURPOSE OF COMPARISON THE OUTPUT OF THE TRANSVERSE DRIVE WAS EQUALIZED TO ZERO AT ZERO DISPLACEMENT 20
FIGURE 3-5 SENSOR STRUCTURE 22 FIGURE 3-6 CLAMPED-CLAMPED FLEXURE 24 FIGURE 3-7 CRAB LEG FLEXURE 25 FIGURE 3-8 FOLDED BEAM FLEXURE 25 FIGURE 3-9 COMPARISON OF THE NON-LINEAR DEFLECTION AMONG THE THREE MICRO-FLEXURES
CONSIDERED CLAMPED-CLAMPED CRAB-LEG AND FOLDED BEAM ALL FLEXURES WERE DESIGNED TO HAVE THE SAME STIFFNESS OF 035 N M-1 26
FIGURE 3-10 LUMPED MASS APPROXIMATION WITH BASE EXCITATION YB(T) 27 FIGURE 3-11 LUMPED MASS APPROXIMATION WITH DIRECT MASS EXCITATION Z(T) 28 FIGURE 3-12 GRAPH REPRESENTING STABILITY REGIONS FOR GAP CLOSING CAPACITIVE DRIVES lsquorsquo
REPRESENTS THE RHS OF EQ (32) FOR T2 WHILE lsquorsquo REPRESENTS THE RHS OF T1 34 FIGURE 4-1 IDEAL CHARGE AMPLIFIER 36 FIGURE 4-2 CAPACITANCE TO VOLTAGE READOUT CIRCUIT WHICH IS MADE UP OF FOUR ELEMENTS THE
CHARGE AMPLIFIER VOLTAGE AMPLIFIER DEMODULATOR AND LOW PASS FILTER 37 FIGURE 4-3 ORCAD SIMULATION RESULTS OF THE CAPACITANCE TO VOLTAGE READOUT CIRCUIT THE
EXCITATION VOLTAGE VE IS 75 VPK AND THE OUTPUT OF THE VOLTAGE AMPLIFIER VVA IS APPROXIMATELY 13125 VPK 38
FIGURE 4-4 ORCAD SIMULATION RESULTS SHOWING THE SINUSOIDAL OUTPUT OF THE DEMODULATOR VDM WITH A 984 VPK-PK AMPLITUDE AND THE DC OUTPUT OF THE LOW PASS FILTER VLP AT APPROXIMATELY 5 V 39
FIGURE 5-1 ORCAD SIMULATION OF THE NOISE VOLTAGE SPECTRAL DENSITY ENO AT THE OUTPUT OF THE VOLTAGE AMPLIFIER 42
FIGURE 5-2 TOTAL OUTPUT NOISE ENO AT THE VOLTAGE AMPLIFIER WHICH IS EXPRESSED AS ENO=( int ENO2)12
42 FIGURE 6-1 POLYMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 44 FIGURE 6-2 SOIMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 45 FIGURE 7-1 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE VERSUS DISPLACEMENT
RELATION WHILE THE SOLID POINTS REPRESENT THE EXPERIMENTAL VALUES THE DASHED LINE REPRESENTS THE ASSUMED LINEAR RELATION FOR DISPLACEMENTS LESS THAN 1ΜM THE EXPERIMENTAL AVERAGE DRIVE SENSITIVITY IS APPROXIMATELY 0024 PF ΜM-1 FOR DISPLACEMENTS GREATER THAN 12 ΜM DRIVE INSTABILITY IS OBSERVED AND IS ATTRIBUTED TO THE PULL-IN EFFECT 47
FIGURE 7-2 lsquotimesrsquo REPRESENTS THE AMPLITUDE AND SOLID LINE REPRESENTS THE PHASE RESPONSE THE RESONANCE PEAK IS LOCATED AT 3600 HZ AND ACCOMPANIED BY A PHASE SHIFT OF 168deg 48
FIGURE 7-3 A SEM MICRO-GRAPH SHOWING THE SENSOR TI FABRICATED USING POLYMUMPS THE FOUR FOLDED BEAM FLEXURES HAD A LENGTH OF 650 ΜM AND TOTAL STIFFNESS OF 14 NM THE EFFECTIVE PROOF MASS WAS 29E-9 KG AND THE SYSTEM HAD A NATURAL FREQUENCY OF 3500 HZ 50
FIGURE 7-4 A SEM MICRO-GRAPH OF THE TRANSVERSE COMB DRIVE IMPLEMENTED IN T1 WITH 200 COMB FINGERS A GAP DISTANCE OF 2 ΜM OVERLAP LENGTH OF 45 ΜM AND THICKNESS OF 2 ΜM THE
vi
AVERAGE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 0024 PF ΜM-1 FOR DISPLACEMENTS OF LESS THAN 1 ΜM 50
FIGURE 7-5 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE TO VOLTAGE RELATION EXPRESSED BY EQ (37) WITH A SENSITIVITY OF 141 V PF-1 THE SOLID POINTS REPRESENT THE EXPERIMENTAL DATA POINTS WHEN DISCRETE CAPACITORS 1-4 PF ARE USED IN THE READOUT CIRCUIT THE HORIZONTAL DASHED LINE REPRESENTS THE OUTPUT VOLTAGE OF THE MEMS SENSOR AT REST AND CORRESPONDS TO A CAPACITANCE OF 35 PF 52
FIGURE 7-6 OSCILLOSCOPE TRACE OF THE INPUT EXCITATION VOLTAGE VE asymp 75 VPK AND OUTPUT OF THE VOLTAGE AMPLIFIER VVA asymp 103 VPK READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-7 OSCILLOSCOPE TRACE OF THE DEMODULATOR OUTPUT VDM asymp 76 VPK-PK AND OUTPUT OF THE LOW PASS FILTER VLP asymp 36 V READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-8 lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS WITH THE DASHED LINE REPRESENTING THE FITTED CURVE SOLID LINE REPRESENTS THE THEORETICAL DATA POINTS ACCORDING TO EQ (10) 54
FIGURE 7-9 EXPERIMENT SETUP WHICH INCLUDES THE MEMS SENOR AND READOUT BOARD MOUNTED ON THE CLAMPED-FREE BEAM AND TEST EQUIPMENT 55
FIGURE 7-10 CLOSE-UP OF THE MEMS SENSOR AND READOUT BOARD MOUNTED ON THE TEST BEAM ALONG WITH THE COMMERCIAL PIEZOELECTRIC ACCELEROMETER ONLY THE CHARGE AMPLIFIER (CA) AND THE VOLTAGE AMPLIFIER (VA) ARE MOUNTED ON THE SAME PROTOTYPE BOARD AS THE MEMS SENSOR DUE TO SPACE CONSTRAINTS 55
FIGURE 7-11 FIGURE COMPARING THE RESPONSE OF A COMMERCIAL PIEZOELECTRIC ACCELEROMETER AND MEMS SENSOR (L1) TO A BEAM VIBRATION A) DYNAMIC RESPONSE OF THE PIEZOELECTRIC ACCELEROMETER REPRESENTING THE TRUE BEAM VIBRATION WITH A PEAK VOLTAGE 114 V CORRESPONDING TO AN APPLIED ACCELERATION OF 114 G B) DYNAMIC RESPONSE OF THE MEMS SENSOR WITH A PEAK VOLTAGE OF 006 V 56
FIGURE 7-12 SENSOR DYNAMIC RESPONSE TO THE VIBRATION OF THE CLAMPED-FREE BEAM AND REPRESENTS A DIRECT EXPORT FROM THE DYNAMIC SIGNAL ANALYZER (AGILENT 35670) TOP TRACE IS THE TIME DOMAIN SIGNAL OF THE LOW PASS FILTER OUTPUT WHERE VLPMAXasymp73MV WHICH REPRESENTS A MAXIMUM ACCELERATION OF APPROXIMATELY 15G THE BOTTOM TRACE REPRESENTS THE FREQUENCY SPECTRUM WITH A PEAK AT 23HZ WHICH CORRESPONDS TO THE BEAM NATURAL FREQUENCY 57
FIGURE 7-13 EXPERIMENTAL RESULTS SHOWING THE RESPONSE OF THE SENSOR TO VARIOUS ACCELERATIONS IN THE RANGE OF 0-20 G WHERE lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS AND THE SOLID LINE REPRESENTS THE FITTED CURVE THE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 4258 MV G-1 58
vii
NOMENCLATURE a applied acceleration A applied acceleration amplitude App parallel plate overlapping area Asldie-film effective area for slide-film damping c coefficient of damping Cf feedback capacitance Cpp capacitance of the parallel plate capacitor cslide-film coefficient of damping as a result of slide-film damping
csqueeze-film coefficient of damping as a result of squeeze-film damping CTotal total capacitance of a comb drive CTotalBottom Total capacitance of the bottom drive in a tri-plate comb drive CTotalL total capacitance of a lateral drive CTotalT total capacitance of a transverse drive CTotalTop Total capacitance of the top drive in a tri-plate comb drive d distance between the parallel plates E Youngrsquos Modulus of sensor material eno noise voltage spectral density Eno total output noise voltage EnoAMP total output amplifier noise voltage EnoR total output resistor noise voltage
f excitation frequency in Hz fn system natural frequency in Hz
Fact electrostatic actuation force
FN noise force Fx-Elect total electrostatic force for pull-in instability Itruss second moment of inertia of truss section on the folded beam flexure Ibeam second moment of inertia of beam section on the folded beam
flexure kair dielectric constant of air at atmospheric pressure kB Boltzmannrsquos constant kT total stiffness of parallel folded beam flexure kx stiffness of various flexures in x-direction ky stiffness of various flexures in y-direction L flexure beam length Lplate effective plate length
viii
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
ACKNOWLEDGMENT
I would first like to thank God for all his blessings and express my gratitude to my
parents and brother for their endless support love and encouragement
I would like to thank my supervisors Dr William Cleghorn and Dr James Mills for
their guidance I would also like to thank my friends and colleagues Henry Chu Dr
Pezhman Hassanpour Shael Markin Dr Lidai Wang and Dr Xuping Zhang at the
Laboratory for Nonlinear Systems Control for great conversation guidance and
friendship Thank you to the members of my Examination Committee Dr Foued Ben
Amara and Dr Jean Zu for their time and feedback
Finally I would like to extend my appreciation to CMC Microsystems for the MEMS
device fabrication and design support and to Autorsquo21 a Network of Federal Centers of
Excellence for project funding
iii
TABLE OF CONTENTS 1 INTRODUCTION 1
11 RESEARCH PURPOSE 1 12 MICROELECTROMECHANICAL SENSORS 2 13 AUTOMOTIVE SENSORS 3 14 OPERATING ENVIRONMENT 4 15 OBJECTIVES AND CONTRIBUTION 5 16 RESEARCH OUTLINE 5
2 LITERATURE REVIEW 7
21 EXISTING METHODS FOR FAULT DETECTION 7 211 Motor Current Monitoring 7 212 Temperature Sensors 8 213 Acoustic Emission Sensors 9
22 MEMS VIBRATION SENSORS 10 221 Resonant Sensors 10 222 Piezoelectric Sensors 11 223 Displacement Variation Sensors 12
3 SENSOR STRUCTURE 15
31 SENSING 15 32 MECHANICAL ANALYSIS 21
321 Micro-Flexure Selection 21 322 Equations of Motion 26 323 Quality Factor 30
33 DRIVE STABILITY 32 34 ELECTROSTATIC ACTUATION 34 35 SECTION SUMMARY 35
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT 36
5 NOISE ANALYSIS 40
51 MECHANICAL-THERMAL NOISE 40 52 ELECTRICAL NOISE 41
6 FABRICATION OVERVIEW 43
61 POLYMUMPS 43 62 SOIMUMPS 44
7 EXPERIMENTAL RESULTS 46
71 MEMS SENSOR ndash T1 46 72 READOUT CIRCUITRY 51 73 MEMS SENSORndash L1 52 74 SECTION SUMMARY 59
8 CONCLUSION AND FUTURE WORK 61
81 CONCLUSION 61 82 FUTURE WORK 62
iv
LIST OF TABLES TABLE 3-1 SPECIFICATIONS FOR EACH CAPACITIVE COMB DRIVE IMPLEMENTED IN THIS WORK T1 L1 AND
T2 21 TABLE 3-2 DIMENSIONS AND MECHANICAL PROPERTIES FOR THE THREE MEMS VIBRATION SENSORS
OUTLINED IN THIS WORK 30 TABLE 3-3 SUMMARY OF THE THEORETICAL COEFFICIENTS OF DAMPING AND QUALITY FACTORS FOR THE
THREE SENSORS PRESENTED IN THIS WORK 32 TABLE 3-4 SUMMARY OF THE CHARACTERISTICS OF THE SENSORS OUTLINED IN THIS WORK 35 TABLE 7-1 SUMMARY OF EXPERIMENTAL RESULTS 60
v
LIST OF FIGURES
FIGURE 3-1 AN INTER-DIGITATED COMB DRIVE WHERE CHANGES IN CAPACITANCE ARE GENERATED BY
EITHER CHANGES IN GAP DISTANCE X0 OR IN THE OVERLAP AREA Y0timesT 16 FIGURE 3-2 AN INTER-DIGITATED COMB DRIVE IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT WHERE X01
X02 AND X03 ARE THE GAP DISTANCES AND X01= X02ltlt X03 TO IMPROVE SENSITIVITY 18 FIGURE 3-3 COMPARISON OF THE EFFECT OF GAP DISTANCES ON LINEARITY FOR TWO CASE lsquorsquo REPRESENTS
CASE 1 WHERE THE FINGERS IN ARE IMPLEMENTED IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT AND X01= X02= 2 ΜM AND X03= 15 ΜM lsquorsquo REPRESENTS THE CASE 2 WHERE THE FINGERS ARE IMPLEMENTED IN A SINGLE ENDED ARRANGEMENT AND X01= X02=X03=2 ΜM 19
FIGURE 3-4 lsquorsquo REPRESENTS THE OUTPUT OF THE DIFFERENTIAL TRI-PLATE DRIVE WHILE lsquorsquo REPRESENTS THE OUTPUT OF THE TRANSVERSE DRIVE THE DASHED LINE REPRESENTS THE LINEAR APPROXIMATION FOR DISPLACEMENTS LESS THAT 1 ΜM FOR THE PURPOSE OF COMPARISON THE OUTPUT OF THE TRANSVERSE DRIVE WAS EQUALIZED TO ZERO AT ZERO DISPLACEMENT 20
FIGURE 3-5 SENSOR STRUCTURE 22 FIGURE 3-6 CLAMPED-CLAMPED FLEXURE 24 FIGURE 3-7 CRAB LEG FLEXURE 25 FIGURE 3-8 FOLDED BEAM FLEXURE 25 FIGURE 3-9 COMPARISON OF THE NON-LINEAR DEFLECTION AMONG THE THREE MICRO-FLEXURES
CONSIDERED CLAMPED-CLAMPED CRAB-LEG AND FOLDED BEAM ALL FLEXURES WERE DESIGNED TO HAVE THE SAME STIFFNESS OF 035 N M-1 26
FIGURE 3-10 LUMPED MASS APPROXIMATION WITH BASE EXCITATION YB(T) 27 FIGURE 3-11 LUMPED MASS APPROXIMATION WITH DIRECT MASS EXCITATION Z(T) 28 FIGURE 3-12 GRAPH REPRESENTING STABILITY REGIONS FOR GAP CLOSING CAPACITIVE DRIVES lsquorsquo
REPRESENTS THE RHS OF EQ (32) FOR T2 WHILE lsquorsquo REPRESENTS THE RHS OF T1 34 FIGURE 4-1 IDEAL CHARGE AMPLIFIER 36 FIGURE 4-2 CAPACITANCE TO VOLTAGE READOUT CIRCUIT WHICH IS MADE UP OF FOUR ELEMENTS THE
CHARGE AMPLIFIER VOLTAGE AMPLIFIER DEMODULATOR AND LOW PASS FILTER 37 FIGURE 4-3 ORCAD SIMULATION RESULTS OF THE CAPACITANCE TO VOLTAGE READOUT CIRCUIT THE
EXCITATION VOLTAGE VE IS 75 VPK AND THE OUTPUT OF THE VOLTAGE AMPLIFIER VVA IS APPROXIMATELY 13125 VPK 38
FIGURE 4-4 ORCAD SIMULATION RESULTS SHOWING THE SINUSOIDAL OUTPUT OF THE DEMODULATOR VDM WITH A 984 VPK-PK AMPLITUDE AND THE DC OUTPUT OF THE LOW PASS FILTER VLP AT APPROXIMATELY 5 V 39
FIGURE 5-1 ORCAD SIMULATION OF THE NOISE VOLTAGE SPECTRAL DENSITY ENO AT THE OUTPUT OF THE VOLTAGE AMPLIFIER 42
FIGURE 5-2 TOTAL OUTPUT NOISE ENO AT THE VOLTAGE AMPLIFIER WHICH IS EXPRESSED AS ENO=( int ENO2)12
42 FIGURE 6-1 POLYMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 44 FIGURE 6-2 SOIMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 45 FIGURE 7-1 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE VERSUS DISPLACEMENT
RELATION WHILE THE SOLID POINTS REPRESENT THE EXPERIMENTAL VALUES THE DASHED LINE REPRESENTS THE ASSUMED LINEAR RELATION FOR DISPLACEMENTS LESS THAN 1ΜM THE EXPERIMENTAL AVERAGE DRIVE SENSITIVITY IS APPROXIMATELY 0024 PF ΜM-1 FOR DISPLACEMENTS GREATER THAN 12 ΜM DRIVE INSTABILITY IS OBSERVED AND IS ATTRIBUTED TO THE PULL-IN EFFECT 47
FIGURE 7-2 lsquotimesrsquo REPRESENTS THE AMPLITUDE AND SOLID LINE REPRESENTS THE PHASE RESPONSE THE RESONANCE PEAK IS LOCATED AT 3600 HZ AND ACCOMPANIED BY A PHASE SHIFT OF 168deg 48
FIGURE 7-3 A SEM MICRO-GRAPH SHOWING THE SENSOR TI FABRICATED USING POLYMUMPS THE FOUR FOLDED BEAM FLEXURES HAD A LENGTH OF 650 ΜM AND TOTAL STIFFNESS OF 14 NM THE EFFECTIVE PROOF MASS WAS 29E-9 KG AND THE SYSTEM HAD A NATURAL FREQUENCY OF 3500 HZ 50
FIGURE 7-4 A SEM MICRO-GRAPH OF THE TRANSVERSE COMB DRIVE IMPLEMENTED IN T1 WITH 200 COMB FINGERS A GAP DISTANCE OF 2 ΜM OVERLAP LENGTH OF 45 ΜM AND THICKNESS OF 2 ΜM THE
vi
AVERAGE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 0024 PF ΜM-1 FOR DISPLACEMENTS OF LESS THAN 1 ΜM 50
FIGURE 7-5 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE TO VOLTAGE RELATION EXPRESSED BY EQ (37) WITH A SENSITIVITY OF 141 V PF-1 THE SOLID POINTS REPRESENT THE EXPERIMENTAL DATA POINTS WHEN DISCRETE CAPACITORS 1-4 PF ARE USED IN THE READOUT CIRCUIT THE HORIZONTAL DASHED LINE REPRESENTS THE OUTPUT VOLTAGE OF THE MEMS SENSOR AT REST AND CORRESPONDS TO A CAPACITANCE OF 35 PF 52
FIGURE 7-6 OSCILLOSCOPE TRACE OF THE INPUT EXCITATION VOLTAGE VE asymp 75 VPK AND OUTPUT OF THE VOLTAGE AMPLIFIER VVA asymp 103 VPK READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-7 OSCILLOSCOPE TRACE OF THE DEMODULATOR OUTPUT VDM asymp 76 VPK-PK AND OUTPUT OF THE LOW PASS FILTER VLP asymp 36 V READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-8 lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS WITH THE DASHED LINE REPRESENTING THE FITTED CURVE SOLID LINE REPRESENTS THE THEORETICAL DATA POINTS ACCORDING TO EQ (10) 54
FIGURE 7-9 EXPERIMENT SETUP WHICH INCLUDES THE MEMS SENOR AND READOUT BOARD MOUNTED ON THE CLAMPED-FREE BEAM AND TEST EQUIPMENT 55
FIGURE 7-10 CLOSE-UP OF THE MEMS SENSOR AND READOUT BOARD MOUNTED ON THE TEST BEAM ALONG WITH THE COMMERCIAL PIEZOELECTRIC ACCELEROMETER ONLY THE CHARGE AMPLIFIER (CA) AND THE VOLTAGE AMPLIFIER (VA) ARE MOUNTED ON THE SAME PROTOTYPE BOARD AS THE MEMS SENSOR DUE TO SPACE CONSTRAINTS 55
FIGURE 7-11 FIGURE COMPARING THE RESPONSE OF A COMMERCIAL PIEZOELECTRIC ACCELEROMETER AND MEMS SENSOR (L1) TO A BEAM VIBRATION A) DYNAMIC RESPONSE OF THE PIEZOELECTRIC ACCELEROMETER REPRESENTING THE TRUE BEAM VIBRATION WITH A PEAK VOLTAGE 114 V CORRESPONDING TO AN APPLIED ACCELERATION OF 114 G B) DYNAMIC RESPONSE OF THE MEMS SENSOR WITH A PEAK VOLTAGE OF 006 V 56
FIGURE 7-12 SENSOR DYNAMIC RESPONSE TO THE VIBRATION OF THE CLAMPED-FREE BEAM AND REPRESENTS A DIRECT EXPORT FROM THE DYNAMIC SIGNAL ANALYZER (AGILENT 35670) TOP TRACE IS THE TIME DOMAIN SIGNAL OF THE LOW PASS FILTER OUTPUT WHERE VLPMAXasymp73MV WHICH REPRESENTS A MAXIMUM ACCELERATION OF APPROXIMATELY 15G THE BOTTOM TRACE REPRESENTS THE FREQUENCY SPECTRUM WITH A PEAK AT 23HZ WHICH CORRESPONDS TO THE BEAM NATURAL FREQUENCY 57
FIGURE 7-13 EXPERIMENTAL RESULTS SHOWING THE RESPONSE OF THE SENSOR TO VARIOUS ACCELERATIONS IN THE RANGE OF 0-20 G WHERE lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS AND THE SOLID LINE REPRESENTS THE FITTED CURVE THE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 4258 MV G-1 58
vii
NOMENCLATURE a applied acceleration A applied acceleration amplitude App parallel plate overlapping area Asldie-film effective area for slide-film damping c coefficient of damping Cf feedback capacitance Cpp capacitance of the parallel plate capacitor cslide-film coefficient of damping as a result of slide-film damping
csqueeze-film coefficient of damping as a result of squeeze-film damping CTotal total capacitance of a comb drive CTotalBottom Total capacitance of the bottom drive in a tri-plate comb drive CTotalL total capacitance of a lateral drive CTotalT total capacitance of a transverse drive CTotalTop Total capacitance of the top drive in a tri-plate comb drive d distance between the parallel plates E Youngrsquos Modulus of sensor material eno noise voltage spectral density Eno total output noise voltage EnoAMP total output amplifier noise voltage EnoR total output resistor noise voltage
f excitation frequency in Hz fn system natural frequency in Hz
Fact electrostatic actuation force
FN noise force Fx-Elect total electrostatic force for pull-in instability Itruss second moment of inertia of truss section on the folded beam flexure Ibeam second moment of inertia of beam section on the folded beam
flexure kair dielectric constant of air at atmospheric pressure kB Boltzmannrsquos constant kT total stiffness of parallel folded beam flexure kx stiffness of various flexures in x-direction ky stiffness of various flexures in y-direction L flexure beam length Lplate effective plate length
viii
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
TABLE OF CONTENTS 1 INTRODUCTION 1
11 RESEARCH PURPOSE 1 12 MICROELECTROMECHANICAL SENSORS 2 13 AUTOMOTIVE SENSORS 3 14 OPERATING ENVIRONMENT 4 15 OBJECTIVES AND CONTRIBUTION 5 16 RESEARCH OUTLINE 5
2 LITERATURE REVIEW 7
21 EXISTING METHODS FOR FAULT DETECTION 7 211 Motor Current Monitoring 7 212 Temperature Sensors 8 213 Acoustic Emission Sensors 9
22 MEMS VIBRATION SENSORS 10 221 Resonant Sensors 10 222 Piezoelectric Sensors 11 223 Displacement Variation Sensors 12
3 SENSOR STRUCTURE 15
31 SENSING 15 32 MECHANICAL ANALYSIS 21
321 Micro-Flexure Selection 21 322 Equations of Motion 26 323 Quality Factor 30
33 DRIVE STABILITY 32 34 ELECTROSTATIC ACTUATION 34 35 SECTION SUMMARY 35
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT 36
5 NOISE ANALYSIS 40
51 MECHANICAL-THERMAL NOISE 40 52 ELECTRICAL NOISE 41
6 FABRICATION OVERVIEW 43
61 POLYMUMPS 43 62 SOIMUMPS 44
7 EXPERIMENTAL RESULTS 46
71 MEMS SENSOR ndash T1 46 72 READOUT CIRCUITRY 51 73 MEMS SENSORndash L1 52 74 SECTION SUMMARY 59
8 CONCLUSION AND FUTURE WORK 61
81 CONCLUSION 61 82 FUTURE WORK 62
iv
LIST OF TABLES TABLE 3-1 SPECIFICATIONS FOR EACH CAPACITIVE COMB DRIVE IMPLEMENTED IN THIS WORK T1 L1 AND
T2 21 TABLE 3-2 DIMENSIONS AND MECHANICAL PROPERTIES FOR THE THREE MEMS VIBRATION SENSORS
OUTLINED IN THIS WORK 30 TABLE 3-3 SUMMARY OF THE THEORETICAL COEFFICIENTS OF DAMPING AND QUALITY FACTORS FOR THE
THREE SENSORS PRESENTED IN THIS WORK 32 TABLE 3-4 SUMMARY OF THE CHARACTERISTICS OF THE SENSORS OUTLINED IN THIS WORK 35 TABLE 7-1 SUMMARY OF EXPERIMENTAL RESULTS 60
v
LIST OF FIGURES
FIGURE 3-1 AN INTER-DIGITATED COMB DRIVE WHERE CHANGES IN CAPACITANCE ARE GENERATED BY
EITHER CHANGES IN GAP DISTANCE X0 OR IN THE OVERLAP AREA Y0timesT 16 FIGURE 3-2 AN INTER-DIGITATED COMB DRIVE IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT WHERE X01
X02 AND X03 ARE THE GAP DISTANCES AND X01= X02ltlt X03 TO IMPROVE SENSITIVITY 18 FIGURE 3-3 COMPARISON OF THE EFFECT OF GAP DISTANCES ON LINEARITY FOR TWO CASE lsquorsquo REPRESENTS
CASE 1 WHERE THE FINGERS IN ARE IMPLEMENTED IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT AND X01= X02= 2 ΜM AND X03= 15 ΜM lsquorsquo REPRESENTS THE CASE 2 WHERE THE FINGERS ARE IMPLEMENTED IN A SINGLE ENDED ARRANGEMENT AND X01= X02=X03=2 ΜM 19
FIGURE 3-4 lsquorsquo REPRESENTS THE OUTPUT OF THE DIFFERENTIAL TRI-PLATE DRIVE WHILE lsquorsquo REPRESENTS THE OUTPUT OF THE TRANSVERSE DRIVE THE DASHED LINE REPRESENTS THE LINEAR APPROXIMATION FOR DISPLACEMENTS LESS THAT 1 ΜM FOR THE PURPOSE OF COMPARISON THE OUTPUT OF THE TRANSVERSE DRIVE WAS EQUALIZED TO ZERO AT ZERO DISPLACEMENT 20
FIGURE 3-5 SENSOR STRUCTURE 22 FIGURE 3-6 CLAMPED-CLAMPED FLEXURE 24 FIGURE 3-7 CRAB LEG FLEXURE 25 FIGURE 3-8 FOLDED BEAM FLEXURE 25 FIGURE 3-9 COMPARISON OF THE NON-LINEAR DEFLECTION AMONG THE THREE MICRO-FLEXURES
CONSIDERED CLAMPED-CLAMPED CRAB-LEG AND FOLDED BEAM ALL FLEXURES WERE DESIGNED TO HAVE THE SAME STIFFNESS OF 035 N M-1 26
FIGURE 3-10 LUMPED MASS APPROXIMATION WITH BASE EXCITATION YB(T) 27 FIGURE 3-11 LUMPED MASS APPROXIMATION WITH DIRECT MASS EXCITATION Z(T) 28 FIGURE 3-12 GRAPH REPRESENTING STABILITY REGIONS FOR GAP CLOSING CAPACITIVE DRIVES lsquorsquo
REPRESENTS THE RHS OF EQ (32) FOR T2 WHILE lsquorsquo REPRESENTS THE RHS OF T1 34 FIGURE 4-1 IDEAL CHARGE AMPLIFIER 36 FIGURE 4-2 CAPACITANCE TO VOLTAGE READOUT CIRCUIT WHICH IS MADE UP OF FOUR ELEMENTS THE
CHARGE AMPLIFIER VOLTAGE AMPLIFIER DEMODULATOR AND LOW PASS FILTER 37 FIGURE 4-3 ORCAD SIMULATION RESULTS OF THE CAPACITANCE TO VOLTAGE READOUT CIRCUIT THE
EXCITATION VOLTAGE VE IS 75 VPK AND THE OUTPUT OF THE VOLTAGE AMPLIFIER VVA IS APPROXIMATELY 13125 VPK 38
FIGURE 4-4 ORCAD SIMULATION RESULTS SHOWING THE SINUSOIDAL OUTPUT OF THE DEMODULATOR VDM WITH A 984 VPK-PK AMPLITUDE AND THE DC OUTPUT OF THE LOW PASS FILTER VLP AT APPROXIMATELY 5 V 39
FIGURE 5-1 ORCAD SIMULATION OF THE NOISE VOLTAGE SPECTRAL DENSITY ENO AT THE OUTPUT OF THE VOLTAGE AMPLIFIER 42
FIGURE 5-2 TOTAL OUTPUT NOISE ENO AT THE VOLTAGE AMPLIFIER WHICH IS EXPRESSED AS ENO=( int ENO2)12
42 FIGURE 6-1 POLYMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 44 FIGURE 6-2 SOIMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 45 FIGURE 7-1 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE VERSUS DISPLACEMENT
RELATION WHILE THE SOLID POINTS REPRESENT THE EXPERIMENTAL VALUES THE DASHED LINE REPRESENTS THE ASSUMED LINEAR RELATION FOR DISPLACEMENTS LESS THAN 1ΜM THE EXPERIMENTAL AVERAGE DRIVE SENSITIVITY IS APPROXIMATELY 0024 PF ΜM-1 FOR DISPLACEMENTS GREATER THAN 12 ΜM DRIVE INSTABILITY IS OBSERVED AND IS ATTRIBUTED TO THE PULL-IN EFFECT 47
FIGURE 7-2 lsquotimesrsquo REPRESENTS THE AMPLITUDE AND SOLID LINE REPRESENTS THE PHASE RESPONSE THE RESONANCE PEAK IS LOCATED AT 3600 HZ AND ACCOMPANIED BY A PHASE SHIFT OF 168deg 48
FIGURE 7-3 A SEM MICRO-GRAPH SHOWING THE SENSOR TI FABRICATED USING POLYMUMPS THE FOUR FOLDED BEAM FLEXURES HAD A LENGTH OF 650 ΜM AND TOTAL STIFFNESS OF 14 NM THE EFFECTIVE PROOF MASS WAS 29E-9 KG AND THE SYSTEM HAD A NATURAL FREQUENCY OF 3500 HZ 50
FIGURE 7-4 A SEM MICRO-GRAPH OF THE TRANSVERSE COMB DRIVE IMPLEMENTED IN T1 WITH 200 COMB FINGERS A GAP DISTANCE OF 2 ΜM OVERLAP LENGTH OF 45 ΜM AND THICKNESS OF 2 ΜM THE
vi
AVERAGE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 0024 PF ΜM-1 FOR DISPLACEMENTS OF LESS THAN 1 ΜM 50
FIGURE 7-5 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE TO VOLTAGE RELATION EXPRESSED BY EQ (37) WITH A SENSITIVITY OF 141 V PF-1 THE SOLID POINTS REPRESENT THE EXPERIMENTAL DATA POINTS WHEN DISCRETE CAPACITORS 1-4 PF ARE USED IN THE READOUT CIRCUIT THE HORIZONTAL DASHED LINE REPRESENTS THE OUTPUT VOLTAGE OF THE MEMS SENSOR AT REST AND CORRESPONDS TO A CAPACITANCE OF 35 PF 52
FIGURE 7-6 OSCILLOSCOPE TRACE OF THE INPUT EXCITATION VOLTAGE VE asymp 75 VPK AND OUTPUT OF THE VOLTAGE AMPLIFIER VVA asymp 103 VPK READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-7 OSCILLOSCOPE TRACE OF THE DEMODULATOR OUTPUT VDM asymp 76 VPK-PK AND OUTPUT OF THE LOW PASS FILTER VLP asymp 36 V READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-8 lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS WITH THE DASHED LINE REPRESENTING THE FITTED CURVE SOLID LINE REPRESENTS THE THEORETICAL DATA POINTS ACCORDING TO EQ (10) 54
FIGURE 7-9 EXPERIMENT SETUP WHICH INCLUDES THE MEMS SENOR AND READOUT BOARD MOUNTED ON THE CLAMPED-FREE BEAM AND TEST EQUIPMENT 55
FIGURE 7-10 CLOSE-UP OF THE MEMS SENSOR AND READOUT BOARD MOUNTED ON THE TEST BEAM ALONG WITH THE COMMERCIAL PIEZOELECTRIC ACCELEROMETER ONLY THE CHARGE AMPLIFIER (CA) AND THE VOLTAGE AMPLIFIER (VA) ARE MOUNTED ON THE SAME PROTOTYPE BOARD AS THE MEMS SENSOR DUE TO SPACE CONSTRAINTS 55
FIGURE 7-11 FIGURE COMPARING THE RESPONSE OF A COMMERCIAL PIEZOELECTRIC ACCELEROMETER AND MEMS SENSOR (L1) TO A BEAM VIBRATION A) DYNAMIC RESPONSE OF THE PIEZOELECTRIC ACCELEROMETER REPRESENTING THE TRUE BEAM VIBRATION WITH A PEAK VOLTAGE 114 V CORRESPONDING TO AN APPLIED ACCELERATION OF 114 G B) DYNAMIC RESPONSE OF THE MEMS SENSOR WITH A PEAK VOLTAGE OF 006 V 56
FIGURE 7-12 SENSOR DYNAMIC RESPONSE TO THE VIBRATION OF THE CLAMPED-FREE BEAM AND REPRESENTS A DIRECT EXPORT FROM THE DYNAMIC SIGNAL ANALYZER (AGILENT 35670) TOP TRACE IS THE TIME DOMAIN SIGNAL OF THE LOW PASS FILTER OUTPUT WHERE VLPMAXasymp73MV WHICH REPRESENTS A MAXIMUM ACCELERATION OF APPROXIMATELY 15G THE BOTTOM TRACE REPRESENTS THE FREQUENCY SPECTRUM WITH A PEAK AT 23HZ WHICH CORRESPONDS TO THE BEAM NATURAL FREQUENCY 57
FIGURE 7-13 EXPERIMENTAL RESULTS SHOWING THE RESPONSE OF THE SENSOR TO VARIOUS ACCELERATIONS IN THE RANGE OF 0-20 G WHERE lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS AND THE SOLID LINE REPRESENTS THE FITTED CURVE THE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 4258 MV G-1 58
vii
NOMENCLATURE a applied acceleration A applied acceleration amplitude App parallel plate overlapping area Asldie-film effective area for slide-film damping c coefficient of damping Cf feedback capacitance Cpp capacitance of the parallel plate capacitor cslide-film coefficient of damping as a result of slide-film damping
csqueeze-film coefficient of damping as a result of squeeze-film damping CTotal total capacitance of a comb drive CTotalBottom Total capacitance of the bottom drive in a tri-plate comb drive CTotalL total capacitance of a lateral drive CTotalT total capacitance of a transverse drive CTotalTop Total capacitance of the top drive in a tri-plate comb drive d distance between the parallel plates E Youngrsquos Modulus of sensor material eno noise voltage spectral density Eno total output noise voltage EnoAMP total output amplifier noise voltage EnoR total output resistor noise voltage
f excitation frequency in Hz fn system natural frequency in Hz
Fact electrostatic actuation force
FN noise force Fx-Elect total electrostatic force for pull-in instability Itruss second moment of inertia of truss section on the folded beam flexure Ibeam second moment of inertia of beam section on the folded beam
flexure kair dielectric constant of air at atmospheric pressure kB Boltzmannrsquos constant kT total stiffness of parallel folded beam flexure kx stiffness of various flexures in x-direction ky stiffness of various flexures in y-direction L flexure beam length Lplate effective plate length
viii
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
LIST OF TABLES TABLE 3-1 SPECIFICATIONS FOR EACH CAPACITIVE COMB DRIVE IMPLEMENTED IN THIS WORK T1 L1 AND
T2 21 TABLE 3-2 DIMENSIONS AND MECHANICAL PROPERTIES FOR THE THREE MEMS VIBRATION SENSORS
OUTLINED IN THIS WORK 30 TABLE 3-3 SUMMARY OF THE THEORETICAL COEFFICIENTS OF DAMPING AND QUALITY FACTORS FOR THE
THREE SENSORS PRESENTED IN THIS WORK 32 TABLE 3-4 SUMMARY OF THE CHARACTERISTICS OF THE SENSORS OUTLINED IN THIS WORK 35 TABLE 7-1 SUMMARY OF EXPERIMENTAL RESULTS 60
v
LIST OF FIGURES
FIGURE 3-1 AN INTER-DIGITATED COMB DRIVE WHERE CHANGES IN CAPACITANCE ARE GENERATED BY
EITHER CHANGES IN GAP DISTANCE X0 OR IN THE OVERLAP AREA Y0timesT 16 FIGURE 3-2 AN INTER-DIGITATED COMB DRIVE IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT WHERE X01
X02 AND X03 ARE THE GAP DISTANCES AND X01= X02ltlt X03 TO IMPROVE SENSITIVITY 18 FIGURE 3-3 COMPARISON OF THE EFFECT OF GAP DISTANCES ON LINEARITY FOR TWO CASE lsquorsquo REPRESENTS
CASE 1 WHERE THE FINGERS IN ARE IMPLEMENTED IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT AND X01= X02= 2 ΜM AND X03= 15 ΜM lsquorsquo REPRESENTS THE CASE 2 WHERE THE FINGERS ARE IMPLEMENTED IN A SINGLE ENDED ARRANGEMENT AND X01= X02=X03=2 ΜM 19
FIGURE 3-4 lsquorsquo REPRESENTS THE OUTPUT OF THE DIFFERENTIAL TRI-PLATE DRIVE WHILE lsquorsquo REPRESENTS THE OUTPUT OF THE TRANSVERSE DRIVE THE DASHED LINE REPRESENTS THE LINEAR APPROXIMATION FOR DISPLACEMENTS LESS THAT 1 ΜM FOR THE PURPOSE OF COMPARISON THE OUTPUT OF THE TRANSVERSE DRIVE WAS EQUALIZED TO ZERO AT ZERO DISPLACEMENT 20
FIGURE 3-5 SENSOR STRUCTURE 22 FIGURE 3-6 CLAMPED-CLAMPED FLEXURE 24 FIGURE 3-7 CRAB LEG FLEXURE 25 FIGURE 3-8 FOLDED BEAM FLEXURE 25 FIGURE 3-9 COMPARISON OF THE NON-LINEAR DEFLECTION AMONG THE THREE MICRO-FLEXURES
CONSIDERED CLAMPED-CLAMPED CRAB-LEG AND FOLDED BEAM ALL FLEXURES WERE DESIGNED TO HAVE THE SAME STIFFNESS OF 035 N M-1 26
FIGURE 3-10 LUMPED MASS APPROXIMATION WITH BASE EXCITATION YB(T) 27 FIGURE 3-11 LUMPED MASS APPROXIMATION WITH DIRECT MASS EXCITATION Z(T) 28 FIGURE 3-12 GRAPH REPRESENTING STABILITY REGIONS FOR GAP CLOSING CAPACITIVE DRIVES lsquorsquo
REPRESENTS THE RHS OF EQ (32) FOR T2 WHILE lsquorsquo REPRESENTS THE RHS OF T1 34 FIGURE 4-1 IDEAL CHARGE AMPLIFIER 36 FIGURE 4-2 CAPACITANCE TO VOLTAGE READOUT CIRCUIT WHICH IS MADE UP OF FOUR ELEMENTS THE
CHARGE AMPLIFIER VOLTAGE AMPLIFIER DEMODULATOR AND LOW PASS FILTER 37 FIGURE 4-3 ORCAD SIMULATION RESULTS OF THE CAPACITANCE TO VOLTAGE READOUT CIRCUIT THE
EXCITATION VOLTAGE VE IS 75 VPK AND THE OUTPUT OF THE VOLTAGE AMPLIFIER VVA IS APPROXIMATELY 13125 VPK 38
FIGURE 4-4 ORCAD SIMULATION RESULTS SHOWING THE SINUSOIDAL OUTPUT OF THE DEMODULATOR VDM WITH A 984 VPK-PK AMPLITUDE AND THE DC OUTPUT OF THE LOW PASS FILTER VLP AT APPROXIMATELY 5 V 39
FIGURE 5-1 ORCAD SIMULATION OF THE NOISE VOLTAGE SPECTRAL DENSITY ENO AT THE OUTPUT OF THE VOLTAGE AMPLIFIER 42
FIGURE 5-2 TOTAL OUTPUT NOISE ENO AT THE VOLTAGE AMPLIFIER WHICH IS EXPRESSED AS ENO=( int ENO2)12
42 FIGURE 6-1 POLYMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 44 FIGURE 6-2 SOIMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 45 FIGURE 7-1 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE VERSUS DISPLACEMENT
RELATION WHILE THE SOLID POINTS REPRESENT THE EXPERIMENTAL VALUES THE DASHED LINE REPRESENTS THE ASSUMED LINEAR RELATION FOR DISPLACEMENTS LESS THAN 1ΜM THE EXPERIMENTAL AVERAGE DRIVE SENSITIVITY IS APPROXIMATELY 0024 PF ΜM-1 FOR DISPLACEMENTS GREATER THAN 12 ΜM DRIVE INSTABILITY IS OBSERVED AND IS ATTRIBUTED TO THE PULL-IN EFFECT 47
FIGURE 7-2 lsquotimesrsquo REPRESENTS THE AMPLITUDE AND SOLID LINE REPRESENTS THE PHASE RESPONSE THE RESONANCE PEAK IS LOCATED AT 3600 HZ AND ACCOMPANIED BY A PHASE SHIFT OF 168deg 48
FIGURE 7-3 A SEM MICRO-GRAPH SHOWING THE SENSOR TI FABRICATED USING POLYMUMPS THE FOUR FOLDED BEAM FLEXURES HAD A LENGTH OF 650 ΜM AND TOTAL STIFFNESS OF 14 NM THE EFFECTIVE PROOF MASS WAS 29E-9 KG AND THE SYSTEM HAD A NATURAL FREQUENCY OF 3500 HZ 50
FIGURE 7-4 A SEM MICRO-GRAPH OF THE TRANSVERSE COMB DRIVE IMPLEMENTED IN T1 WITH 200 COMB FINGERS A GAP DISTANCE OF 2 ΜM OVERLAP LENGTH OF 45 ΜM AND THICKNESS OF 2 ΜM THE
vi
AVERAGE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 0024 PF ΜM-1 FOR DISPLACEMENTS OF LESS THAN 1 ΜM 50
FIGURE 7-5 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE TO VOLTAGE RELATION EXPRESSED BY EQ (37) WITH A SENSITIVITY OF 141 V PF-1 THE SOLID POINTS REPRESENT THE EXPERIMENTAL DATA POINTS WHEN DISCRETE CAPACITORS 1-4 PF ARE USED IN THE READOUT CIRCUIT THE HORIZONTAL DASHED LINE REPRESENTS THE OUTPUT VOLTAGE OF THE MEMS SENSOR AT REST AND CORRESPONDS TO A CAPACITANCE OF 35 PF 52
FIGURE 7-6 OSCILLOSCOPE TRACE OF THE INPUT EXCITATION VOLTAGE VE asymp 75 VPK AND OUTPUT OF THE VOLTAGE AMPLIFIER VVA asymp 103 VPK READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-7 OSCILLOSCOPE TRACE OF THE DEMODULATOR OUTPUT VDM asymp 76 VPK-PK AND OUTPUT OF THE LOW PASS FILTER VLP asymp 36 V READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-8 lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS WITH THE DASHED LINE REPRESENTING THE FITTED CURVE SOLID LINE REPRESENTS THE THEORETICAL DATA POINTS ACCORDING TO EQ (10) 54
FIGURE 7-9 EXPERIMENT SETUP WHICH INCLUDES THE MEMS SENOR AND READOUT BOARD MOUNTED ON THE CLAMPED-FREE BEAM AND TEST EQUIPMENT 55
FIGURE 7-10 CLOSE-UP OF THE MEMS SENSOR AND READOUT BOARD MOUNTED ON THE TEST BEAM ALONG WITH THE COMMERCIAL PIEZOELECTRIC ACCELEROMETER ONLY THE CHARGE AMPLIFIER (CA) AND THE VOLTAGE AMPLIFIER (VA) ARE MOUNTED ON THE SAME PROTOTYPE BOARD AS THE MEMS SENSOR DUE TO SPACE CONSTRAINTS 55
FIGURE 7-11 FIGURE COMPARING THE RESPONSE OF A COMMERCIAL PIEZOELECTRIC ACCELEROMETER AND MEMS SENSOR (L1) TO A BEAM VIBRATION A) DYNAMIC RESPONSE OF THE PIEZOELECTRIC ACCELEROMETER REPRESENTING THE TRUE BEAM VIBRATION WITH A PEAK VOLTAGE 114 V CORRESPONDING TO AN APPLIED ACCELERATION OF 114 G B) DYNAMIC RESPONSE OF THE MEMS SENSOR WITH A PEAK VOLTAGE OF 006 V 56
FIGURE 7-12 SENSOR DYNAMIC RESPONSE TO THE VIBRATION OF THE CLAMPED-FREE BEAM AND REPRESENTS A DIRECT EXPORT FROM THE DYNAMIC SIGNAL ANALYZER (AGILENT 35670) TOP TRACE IS THE TIME DOMAIN SIGNAL OF THE LOW PASS FILTER OUTPUT WHERE VLPMAXasymp73MV WHICH REPRESENTS A MAXIMUM ACCELERATION OF APPROXIMATELY 15G THE BOTTOM TRACE REPRESENTS THE FREQUENCY SPECTRUM WITH A PEAK AT 23HZ WHICH CORRESPONDS TO THE BEAM NATURAL FREQUENCY 57
FIGURE 7-13 EXPERIMENTAL RESULTS SHOWING THE RESPONSE OF THE SENSOR TO VARIOUS ACCELERATIONS IN THE RANGE OF 0-20 G WHERE lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS AND THE SOLID LINE REPRESENTS THE FITTED CURVE THE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 4258 MV G-1 58
vii
NOMENCLATURE a applied acceleration A applied acceleration amplitude App parallel plate overlapping area Asldie-film effective area for slide-film damping c coefficient of damping Cf feedback capacitance Cpp capacitance of the parallel plate capacitor cslide-film coefficient of damping as a result of slide-film damping
csqueeze-film coefficient of damping as a result of squeeze-film damping CTotal total capacitance of a comb drive CTotalBottom Total capacitance of the bottom drive in a tri-plate comb drive CTotalL total capacitance of a lateral drive CTotalT total capacitance of a transverse drive CTotalTop Total capacitance of the top drive in a tri-plate comb drive d distance between the parallel plates E Youngrsquos Modulus of sensor material eno noise voltage spectral density Eno total output noise voltage EnoAMP total output amplifier noise voltage EnoR total output resistor noise voltage
f excitation frequency in Hz fn system natural frequency in Hz
Fact electrostatic actuation force
FN noise force Fx-Elect total electrostatic force for pull-in instability Itruss second moment of inertia of truss section on the folded beam flexure Ibeam second moment of inertia of beam section on the folded beam
flexure kair dielectric constant of air at atmospheric pressure kB Boltzmannrsquos constant kT total stiffness of parallel folded beam flexure kx stiffness of various flexures in x-direction ky stiffness of various flexures in y-direction L flexure beam length Lplate effective plate length
viii
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
LIST OF FIGURES
FIGURE 3-1 AN INTER-DIGITATED COMB DRIVE WHERE CHANGES IN CAPACITANCE ARE GENERATED BY
EITHER CHANGES IN GAP DISTANCE X0 OR IN THE OVERLAP AREA Y0timesT 16 FIGURE 3-2 AN INTER-DIGITATED COMB DRIVE IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT WHERE X01
X02 AND X03 ARE THE GAP DISTANCES AND X01= X02ltlt X03 TO IMPROVE SENSITIVITY 18 FIGURE 3-3 COMPARISON OF THE EFFECT OF GAP DISTANCES ON LINEARITY FOR TWO CASE lsquorsquo REPRESENTS
CASE 1 WHERE THE FINGERS IN ARE IMPLEMENTED IN A DIFFERENTIAL TRI-PLATE ARRANGEMENT AND X01= X02= 2 ΜM AND X03= 15 ΜM lsquorsquo REPRESENTS THE CASE 2 WHERE THE FINGERS ARE IMPLEMENTED IN A SINGLE ENDED ARRANGEMENT AND X01= X02=X03=2 ΜM 19
FIGURE 3-4 lsquorsquo REPRESENTS THE OUTPUT OF THE DIFFERENTIAL TRI-PLATE DRIVE WHILE lsquorsquo REPRESENTS THE OUTPUT OF THE TRANSVERSE DRIVE THE DASHED LINE REPRESENTS THE LINEAR APPROXIMATION FOR DISPLACEMENTS LESS THAT 1 ΜM FOR THE PURPOSE OF COMPARISON THE OUTPUT OF THE TRANSVERSE DRIVE WAS EQUALIZED TO ZERO AT ZERO DISPLACEMENT 20
FIGURE 3-5 SENSOR STRUCTURE 22 FIGURE 3-6 CLAMPED-CLAMPED FLEXURE 24 FIGURE 3-7 CRAB LEG FLEXURE 25 FIGURE 3-8 FOLDED BEAM FLEXURE 25 FIGURE 3-9 COMPARISON OF THE NON-LINEAR DEFLECTION AMONG THE THREE MICRO-FLEXURES
CONSIDERED CLAMPED-CLAMPED CRAB-LEG AND FOLDED BEAM ALL FLEXURES WERE DESIGNED TO HAVE THE SAME STIFFNESS OF 035 N M-1 26
FIGURE 3-10 LUMPED MASS APPROXIMATION WITH BASE EXCITATION YB(T) 27 FIGURE 3-11 LUMPED MASS APPROXIMATION WITH DIRECT MASS EXCITATION Z(T) 28 FIGURE 3-12 GRAPH REPRESENTING STABILITY REGIONS FOR GAP CLOSING CAPACITIVE DRIVES lsquorsquo
REPRESENTS THE RHS OF EQ (32) FOR T2 WHILE lsquorsquo REPRESENTS THE RHS OF T1 34 FIGURE 4-1 IDEAL CHARGE AMPLIFIER 36 FIGURE 4-2 CAPACITANCE TO VOLTAGE READOUT CIRCUIT WHICH IS MADE UP OF FOUR ELEMENTS THE
CHARGE AMPLIFIER VOLTAGE AMPLIFIER DEMODULATOR AND LOW PASS FILTER 37 FIGURE 4-3 ORCAD SIMULATION RESULTS OF THE CAPACITANCE TO VOLTAGE READOUT CIRCUIT THE
EXCITATION VOLTAGE VE IS 75 VPK AND THE OUTPUT OF THE VOLTAGE AMPLIFIER VVA IS APPROXIMATELY 13125 VPK 38
FIGURE 4-4 ORCAD SIMULATION RESULTS SHOWING THE SINUSOIDAL OUTPUT OF THE DEMODULATOR VDM WITH A 984 VPK-PK AMPLITUDE AND THE DC OUTPUT OF THE LOW PASS FILTER VLP AT APPROXIMATELY 5 V 39
FIGURE 5-1 ORCAD SIMULATION OF THE NOISE VOLTAGE SPECTRAL DENSITY ENO AT THE OUTPUT OF THE VOLTAGE AMPLIFIER 42
FIGURE 5-2 TOTAL OUTPUT NOISE ENO AT THE VOLTAGE AMPLIFIER WHICH IS EXPRESSED AS ENO=( int ENO2)12
42 FIGURE 6-1 POLYMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 44 FIGURE 6-2 SOIMUMPS FABRICATION PROCESS FOR A SIMPLE CANTILEVER BEAM 45 FIGURE 7-1 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE VERSUS DISPLACEMENT
RELATION WHILE THE SOLID POINTS REPRESENT THE EXPERIMENTAL VALUES THE DASHED LINE REPRESENTS THE ASSUMED LINEAR RELATION FOR DISPLACEMENTS LESS THAN 1ΜM THE EXPERIMENTAL AVERAGE DRIVE SENSITIVITY IS APPROXIMATELY 0024 PF ΜM-1 FOR DISPLACEMENTS GREATER THAN 12 ΜM DRIVE INSTABILITY IS OBSERVED AND IS ATTRIBUTED TO THE PULL-IN EFFECT 47
FIGURE 7-2 lsquotimesrsquo REPRESENTS THE AMPLITUDE AND SOLID LINE REPRESENTS THE PHASE RESPONSE THE RESONANCE PEAK IS LOCATED AT 3600 HZ AND ACCOMPANIED BY A PHASE SHIFT OF 168deg 48
FIGURE 7-3 A SEM MICRO-GRAPH SHOWING THE SENSOR TI FABRICATED USING POLYMUMPS THE FOUR FOLDED BEAM FLEXURES HAD A LENGTH OF 650 ΜM AND TOTAL STIFFNESS OF 14 NM THE EFFECTIVE PROOF MASS WAS 29E-9 KG AND THE SYSTEM HAD A NATURAL FREQUENCY OF 3500 HZ 50
FIGURE 7-4 A SEM MICRO-GRAPH OF THE TRANSVERSE COMB DRIVE IMPLEMENTED IN T1 WITH 200 COMB FINGERS A GAP DISTANCE OF 2 ΜM OVERLAP LENGTH OF 45 ΜM AND THICKNESS OF 2 ΜM THE
vi
AVERAGE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 0024 PF ΜM-1 FOR DISPLACEMENTS OF LESS THAN 1 ΜM 50
FIGURE 7-5 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE TO VOLTAGE RELATION EXPRESSED BY EQ (37) WITH A SENSITIVITY OF 141 V PF-1 THE SOLID POINTS REPRESENT THE EXPERIMENTAL DATA POINTS WHEN DISCRETE CAPACITORS 1-4 PF ARE USED IN THE READOUT CIRCUIT THE HORIZONTAL DASHED LINE REPRESENTS THE OUTPUT VOLTAGE OF THE MEMS SENSOR AT REST AND CORRESPONDS TO A CAPACITANCE OF 35 PF 52
FIGURE 7-6 OSCILLOSCOPE TRACE OF THE INPUT EXCITATION VOLTAGE VE asymp 75 VPK AND OUTPUT OF THE VOLTAGE AMPLIFIER VVA asymp 103 VPK READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-7 OSCILLOSCOPE TRACE OF THE DEMODULATOR OUTPUT VDM asymp 76 VPK-PK AND OUTPUT OF THE LOW PASS FILTER VLP asymp 36 V READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-8 lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS WITH THE DASHED LINE REPRESENTING THE FITTED CURVE SOLID LINE REPRESENTS THE THEORETICAL DATA POINTS ACCORDING TO EQ (10) 54
FIGURE 7-9 EXPERIMENT SETUP WHICH INCLUDES THE MEMS SENOR AND READOUT BOARD MOUNTED ON THE CLAMPED-FREE BEAM AND TEST EQUIPMENT 55
FIGURE 7-10 CLOSE-UP OF THE MEMS SENSOR AND READOUT BOARD MOUNTED ON THE TEST BEAM ALONG WITH THE COMMERCIAL PIEZOELECTRIC ACCELEROMETER ONLY THE CHARGE AMPLIFIER (CA) AND THE VOLTAGE AMPLIFIER (VA) ARE MOUNTED ON THE SAME PROTOTYPE BOARD AS THE MEMS SENSOR DUE TO SPACE CONSTRAINTS 55
FIGURE 7-11 FIGURE COMPARING THE RESPONSE OF A COMMERCIAL PIEZOELECTRIC ACCELEROMETER AND MEMS SENSOR (L1) TO A BEAM VIBRATION A) DYNAMIC RESPONSE OF THE PIEZOELECTRIC ACCELEROMETER REPRESENTING THE TRUE BEAM VIBRATION WITH A PEAK VOLTAGE 114 V CORRESPONDING TO AN APPLIED ACCELERATION OF 114 G B) DYNAMIC RESPONSE OF THE MEMS SENSOR WITH A PEAK VOLTAGE OF 006 V 56
FIGURE 7-12 SENSOR DYNAMIC RESPONSE TO THE VIBRATION OF THE CLAMPED-FREE BEAM AND REPRESENTS A DIRECT EXPORT FROM THE DYNAMIC SIGNAL ANALYZER (AGILENT 35670) TOP TRACE IS THE TIME DOMAIN SIGNAL OF THE LOW PASS FILTER OUTPUT WHERE VLPMAXasymp73MV WHICH REPRESENTS A MAXIMUM ACCELERATION OF APPROXIMATELY 15G THE BOTTOM TRACE REPRESENTS THE FREQUENCY SPECTRUM WITH A PEAK AT 23HZ WHICH CORRESPONDS TO THE BEAM NATURAL FREQUENCY 57
FIGURE 7-13 EXPERIMENTAL RESULTS SHOWING THE RESPONSE OF THE SENSOR TO VARIOUS ACCELERATIONS IN THE RANGE OF 0-20 G WHERE lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS AND THE SOLID LINE REPRESENTS THE FITTED CURVE THE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 4258 MV G-1 58
vii
NOMENCLATURE a applied acceleration A applied acceleration amplitude App parallel plate overlapping area Asldie-film effective area for slide-film damping c coefficient of damping Cf feedback capacitance Cpp capacitance of the parallel plate capacitor cslide-film coefficient of damping as a result of slide-film damping
csqueeze-film coefficient of damping as a result of squeeze-film damping CTotal total capacitance of a comb drive CTotalBottom Total capacitance of the bottom drive in a tri-plate comb drive CTotalL total capacitance of a lateral drive CTotalT total capacitance of a transverse drive CTotalTop Total capacitance of the top drive in a tri-plate comb drive d distance between the parallel plates E Youngrsquos Modulus of sensor material eno noise voltage spectral density Eno total output noise voltage EnoAMP total output amplifier noise voltage EnoR total output resistor noise voltage
f excitation frequency in Hz fn system natural frequency in Hz
Fact electrostatic actuation force
FN noise force Fx-Elect total electrostatic force for pull-in instability Itruss second moment of inertia of truss section on the folded beam flexure Ibeam second moment of inertia of beam section on the folded beam
flexure kair dielectric constant of air at atmospheric pressure kB Boltzmannrsquos constant kT total stiffness of parallel folded beam flexure kx stiffness of various flexures in x-direction ky stiffness of various flexures in y-direction L flexure beam length Lplate effective plate length
viii
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
AVERAGE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 0024 PF ΜM-1 FOR DISPLACEMENTS OF LESS THAN 1 ΜM 50
FIGURE 7-5 THE SOLID LINE REPRESENTS THE THEORETICAL CAPACITANCE TO VOLTAGE RELATION EXPRESSED BY EQ (37) WITH A SENSITIVITY OF 141 V PF-1 THE SOLID POINTS REPRESENT THE EXPERIMENTAL DATA POINTS WHEN DISCRETE CAPACITORS 1-4 PF ARE USED IN THE READOUT CIRCUIT THE HORIZONTAL DASHED LINE REPRESENTS THE OUTPUT VOLTAGE OF THE MEMS SENSOR AT REST AND CORRESPONDS TO A CAPACITANCE OF 35 PF 52
FIGURE 7-6 OSCILLOSCOPE TRACE OF THE INPUT EXCITATION VOLTAGE VE asymp 75 VPK AND OUTPUT OF THE VOLTAGE AMPLIFIER VVA asymp 103 VPK READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-7 OSCILLOSCOPE TRACE OF THE DEMODULATOR OUTPUT VDM asymp 76 VPK-PK AND OUTPUT OF THE LOW PASS FILTER VLP asymp 36 V READINGS WERE TAKEN WITH THE MEMS SENSOR AT REST 53
FIGURE 7-8 lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS WITH THE DASHED LINE REPRESENTING THE FITTED CURVE SOLID LINE REPRESENTS THE THEORETICAL DATA POINTS ACCORDING TO EQ (10) 54
FIGURE 7-9 EXPERIMENT SETUP WHICH INCLUDES THE MEMS SENOR AND READOUT BOARD MOUNTED ON THE CLAMPED-FREE BEAM AND TEST EQUIPMENT 55
FIGURE 7-10 CLOSE-UP OF THE MEMS SENSOR AND READOUT BOARD MOUNTED ON THE TEST BEAM ALONG WITH THE COMMERCIAL PIEZOELECTRIC ACCELEROMETER ONLY THE CHARGE AMPLIFIER (CA) AND THE VOLTAGE AMPLIFIER (VA) ARE MOUNTED ON THE SAME PROTOTYPE BOARD AS THE MEMS SENSOR DUE TO SPACE CONSTRAINTS 55
FIGURE 7-11 FIGURE COMPARING THE RESPONSE OF A COMMERCIAL PIEZOELECTRIC ACCELEROMETER AND MEMS SENSOR (L1) TO A BEAM VIBRATION A) DYNAMIC RESPONSE OF THE PIEZOELECTRIC ACCELEROMETER REPRESENTING THE TRUE BEAM VIBRATION WITH A PEAK VOLTAGE 114 V CORRESPONDING TO AN APPLIED ACCELERATION OF 114 G B) DYNAMIC RESPONSE OF THE MEMS SENSOR WITH A PEAK VOLTAGE OF 006 V 56
FIGURE 7-12 SENSOR DYNAMIC RESPONSE TO THE VIBRATION OF THE CLAMPED-FREE BEAM AND REPRESENTS A DIRECT EXPORT FROM THE DYNAMIC SIGNAL ANALYZER (AGILENT 35670) TOP TRACE IS THE TIME DOMAIN SIGNAL OF THE LOW PASS FILTER OUTPUT WHERE VLPMAXasymp73MV WHICH REPRESENTS A MAXIMUM ACCELERATION OF APPROXIMATELY 15G THE BOTTOM TRACE REPRESENTS THE FREQUENCY SPECTRUM WITH A PEAK AT 23HZ WHICH CORRESPONDS TO THE BEAM NATURAL FREQUENCY 57
FIGURE 7-13 EXPERIMENTAL RESULTS SHOWING THE RESPONSE OF THE SENSOR TO VARIOUS ACCELERATIONS IN THE RANGE OF 0-20 G WHERE lsquorsquo REPRESENTS THE EXPERIMENTAL DATA POINTS AND THE SOLID LINE REPRESENTS THE FITTED CURVE THE EXPERIMENTAL SENSITIVITY WAS FOUND TO BE 4258 MV G-1 58
vii
NOMENCLATURE a applied acceleration A applied acceleration amplitude App parallel plate overlapping area Asldie-film effective area for slide-film damping c coefficient of damping Cf feedback capacitance Cpp capacitance of the parallel plate capacitor cslide-film coefficient of damping as a result of slide-film damping
csqueeze-film coefficient of damping as a result of squeeze-film damping CTotal total capacitance of a comb drive CTotalBottom Total capacitance of the bottom drive in a tri-plate comb drive CTotalL total capacitance of a lateral drive CTotalT total capacitance of a transverse drive CTotalTop Total capacitance of the top drive in a tri-plate comb drive d distance between the parallel plates E Youngrsquos Modulus of sensor material eno noise voltage spectral density Eno total output noise voltage EnoAMP total output amplifier noise voltage EnoR total output resistor noise voltage
f excitation frequency in Hz fn system natural frequency in Hz
Fact electrostatic actuation force
FN noise force Fx-Elect total electrostatic force for pull-in instability Itruss second moment of inertia of truss section on the folded beam flexure Ibeam second moment of inertia of beam section on the folded beam
flexure kair dielectric constant of air at atmospheric pressure kB Boltzmannrsquos constant kT total stiffness of parallel folded beam flexure kx stiffness of various flexures in x-direction ky stiffness of various flexures in y-direction L flexure beam length Lplate effective plate length
viii
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
NOMENCLATURE a applied acceleration A applied acceleration amplitude App parallel plate overlapping area Asldie-film effective area for slide-film damping c coefficient of damping Cf feedback capacitance Cpp capacitance of the parallel plate capacitor cslide-film coefficient of damping as a result of slide-film damping
csqueeze-film coefficient of damping as a result of squeeze-film damping CTotal total capacitance of a comb drive CTotalBottom Total capacitance of the bottom drive in a tri-plate comb drive CTotalL total capacitance of a lateral drive CTotalT total capacitance of a transverse drive CTotalTop Total capacitance of the top drive in a tri-plate comb drive d distance between the parallel plates E Youngrsquos Modulus of sensor material eno noise voltage spectral density Eno total output noise voltage EnoAMP total output amplifier noise voltage EnoR total output resistor noise voltage
f excitation frequency in Hz fn system natural frequency in Hz
Fact electrostatic actuation force
FN noise force Fx-Elect total electrostatic force for pull-in instability Itruss second moment of inertia of truss section on the folded beam flexure Ibeam second moment of inertia of beam section on the folded beam
flexure kair dielectric constant of air at atmospheric pressure kB Boltzmannrsquos constant kT total stiffness of parallel folded beam flexure kx stiffness of various flexures in x-direction ky stiffness of various flexures in y-direction L flexure beam length Lplate effective plate length
viii
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
Ls shin beam length Lt thigh beam length m effective proof mass N number of comb fingers on the movable comb
NB number of balls
Qtotal total quality factor
Qf energy lost to surrounding fluid
Qs energy lost through supports
Qm energy lost through sensor material S theoretical sensitivity of sensor SL sensitivity of a lateral comb drive SM theoretical mechanical sensitivity of the proof massflexure system SRO sensitivity of readout circuit ST sensitivity of a transverse comb drive STAVG average sensitivity of a transverse comb drive t time T absolute temperature tb flexure beam thickness tc out of plane comb finger thickness
Vact electrostatic actuation voltage VCA output amplitude of charge amplifier VDM output amplitide of demodulator VE input amplitude of sinusoidal excitation voltage VLP output of low pass filter VLPMAX maximum output voltage at low pass filter VVA output amplitude of voltage amplifier vy relative motion of proof mass with respect to base Vy relative motion amplitude w flexure beam width Wplate effective plate width ws shin beam width wt thigh beam width x0 initial comb finger gap
x01 small gap distance on top side of the tri-plate comb drive
x02 small gap distance on bottom side of the tri-plate comb drive
ix
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
x03 larger gap distance on the tri-plate comb drive
XN noise displacement y0 initial comb finger overlap length yB base displacement YB base displacement amplitude ym absolute proof mass displacement α parameter dependant on the WplateLplate ratio ∆C Change in capacitance from a tri-plate drive in a differential
arrangement ε permittivity of free space κ second moment of inertia ratio (shinthigh) ζ damping ratio microair absolute viscosity of air ω base excitation frequency in rads ωn sensor natural frequency in rads ωv electrostatic actuation voltage frequency in rads L1 Sensor fabricated using SoiMUMPS with a lateral comb drive T1 Sensor fabricated using PolyMUMPS with a transverse comb drive T2 Sensor fabricated using SoiMUMPS with a tri-plate comb drive
x
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
1 INTRODUCTION
11 RESEARCH PURPOSE
The modern automobile consists of many mechanical systems such as power seats
windshield wipers mirrors trunks and windows which are all susceptible to breakdown
Without any condition monitoring system the breakdown is usually catastrophic and
requires an expensive part replacement Real-time condition monitoring allows for early
detection of faults which could require a simple solution such as the application of a
lubricant to fix This prolongs the useful life of the component and prevents sudden and
unexpected failure Real-time condition monitoring can be accomplished by examining the
vibration signature of a mechanical system For example an automobile power window
consists of a DC motor and its associated bearings and couplings a gear reduction system
consisting of worm and spur gears and kinematic links Faults resulting in excessive
vibrations may be caused by coupling misalignment bearing failure or gear train failure
Coupling misalignments occur at the connection between the drive shaft and the driven
shaft and are typically due to imperfect manufacturing Bearing failure is caused by lack
of lubrication or moisture contamination causing rusting while gear train failure is caused
by misaligned broken cracked or chipped gear teeth [1] Each fault occurs at a
characteristic frequency and so the state of the mechanical system can be determined by
monitoring the amplitudes of the relevant frequencies Vibrations due to coupling
misalignments occur at harmonics of the shaft rotational speed Gear vibrations occur at
the gear turn speed or at sidebands of the gear mesh frequency [1] Ball bearing vibrations
1
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
2
may be caused by outer bearing race defects fOB inner bearing race defects fIB ball
defects fB and train defects fT which all occur at specific frequencies [12]
cos( )12
dBOB r
p
bNf fd
β = minus
cos( )12
dBIB r
p
bNf fd
β = +
2cos( )1
2p r d
Bd p
d f bfb d
β = minus
cos( )12
drT
p
bffd
β = minus
(1)
where NB is the number of balls fr is the rotation frequency bd is the ball diameter β is the
contact angle between the ball and races and dp is the ball pitch diameter By monitoring
the real time vibration signature of the mechanical system anomalies can be quickly
identified and fixed
12 MICROELECTROMECHANICAL SENSORS
Microelectromechanical systems (MEMS) inertial sensors provide a small footprint
with sensitivities that are either comparable or exceed any macro sensor along with the
capability of mass production and low unit cost These sensors utilize compliant micro-
flexures attached to a proof mass that displaces in response to an environmental
acceleration Many transduction mechanisms have been developed that convert the
displacement into a measurable electric signal and include thermal piezoresistive
piezoelectric optical and capacitive methods Most MEMS sensors are silicon based and
are fabricated using surface or bulk micromachining Surface micromachining creates free
standing movable structures on top of a substrate using a combination of sacrificial layers
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
3
and structural layers which are commonly polysilicon [3] In bulk micromachining the
mechanical structures are defined using a removal process where bulk material typically
silicon is etched away MEMS sensors have been used for vibration and shock monitoring
on industrial systems and robotics guidance and navigation in global positioning systems
(GPS) seismometry in earthquake prediction image stabilization in digital cameras and
automobile safety and stability [4]
13 AUTOMOTIVE SENSORS
Sensors cover every major aspect of a modern automobile power-train sensors
monitor fuel combustion and emissions chassis sensors monitor road traction and tire
condition and body sensors facilitate air-bag deployment and vehicle proximity for radar
guided cruise control [5] Pressure sensors which typically consist of a piezoresistive
strain sensing element attached to a silicon diaphragm that deflects when exposed to an
applied pressure are one of the first micro-machined sensors used in an automobile
Implemented as manifold absolute pressure (MAP) sensors they allow precise control of
the air fuel ratio which allows the catalytic converter to efficiently reduce tailpipe
emissions [6] Variable reluctance sensors based on electromagnetics are used for
automobile traction control and produce a voltage output that is dependant on the
magnetic flux variations between a rotating component and the sensor bias magnet [57]
MEMS based linear accelerometers are utilized for airbag deployment upon impact and
provide a lower cost and smaller package over the older ball-in-tube accelerometers [6]
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
4
14 OPERATING ENVIRONMENT
Through proper packaging MEMS sensors can withstand the harsh automotive
environment where they face cyclic temperatures (-40degC to 150degC) mechanical shock or
vibration (10g-500g) and exposure to various fluids (oil brake fluid or engine coolant)
and airborne particles [8] Temperature changes induce thermal stresses causing
undesirable bending of the silicon microstructures that changes their dynamic response
Mechanical shock which could occur during a high impact collision or vibration from
uneven road conditions cause the delicate microstructures to fracture at points of stress
concentration These effects could be reduced by efficient design Use of a folded beam
micro-flexure that allows expansion and contraction or use of fillets at corners which
reduce the chance of fracture Bonding pads wire bonding and onboard circuit
components can be protected from air borne particles which could cause shorting using
parylene or gels Alternatively the entire package can be sealed hermetically which also
protects the MEMS device [8] In some cases such as fluid pressure sensors intimate
contact between the MEMS device and potentially corrosive fluid is necessary A recent
approach used a nanometer thin coating of silicon carbide (SiC) over a foundry fabricated
MEMS sensor using low temperature low pressure chemical vapor deposition (LPCVD)
[9] The SiC known for its chemical inertness did not compromise the mechanical
characteristics of the original design Therefore MEMS provides a strong platform for an
automotive sensor
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
5
15 OBJECTIVES AND CONTRIBUTION
The objective of this work is the development of a low frequency MEMS vibration
sensor The sensor must be low cost and mass producible which can be achieved using
standard foundry fabrication processes such as the Polysilicon Multi-User MEMS
Processes (PolyMUMPS) or the Silicon-on-Insulator Multi-User MEMS Processes
(SoiMUMPS) In addition the number of additional processing steps which are required
if integrated CMOS circuitry is used or if vacuum packaging is required must be
minimized This can be achieved using an off-chip circuit made with widely available
discrete components and sensor operation in air
In addition this work presents a MEMS sensor that is fabricated using standard
foundry processes such as PolyMUMPS and SoiMUMPS compared to many MEMS
sensors in literature that are fabricated using customized processes
16 RESEARCH OUTLINE
This work demonstrates three iterations of a low frequency capacitive vibration
sensor fabricated using the standard foundry processes Each sensor consists of a
suspended proof mass that displaces in response to an external vibration The proof mass
is attached to a folded beam micro-flexure compliant in one direction and its lateral
movement is sensed using a capacitive comb drive The first sensor T1 is fabricated in
PolyMUMPS and uses a transverse capacitive comb drive The second L1 fabricated in
SoiMUMPS uses a fully linear lateral capacitive drive and achieves an increase in
sensitivity from T1 Finally the third sensor T2 is fabricated in SoiMUMPS and uses a
differential tri-plate capacitive drive This sensor takes advantage of the increase in
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
6
sensitivity gained from a transverse drive but rejects off-axis displacement errors and
common mode using a differential tri-plate arrangement In addition an off-chip
capacitance to voltage readout circuit is fabricated and tested on sensor L1
Chapter 2 begins with an overview on the various methods of fault detection
followed by a review of MEMS vibration sensors each with a different transduction
mechanism Chapter 3 outlines the sensor structure focusing on the mechanical design
and sensing principal Chapter 4 focuses on the readout circuitry layout and simulation
results Chapter 5 outlines the various source of noise including mechanical noise arising
from the sensor and electric noise from the readout circuit Chapter 6 overviews the
foundry fabrication processes used including PolyMUMPs and SoiMUMPS The
experimental results of two sensors T1 and L1 which were fabricated and tested within
the thesis time frame are presented in Chapter 7 Chapter 8 concludes the thesis with a
brief summary and outline for future work
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
2 LITERATURE REVIEW
21 EXISTING METHODS FOR FAULT DETECTION
As this thesis develops a sensor for fault detection using vibration monitoring this
section overviews three existing methods for fault detection in mechanical systems First
motor current monitoring is discussed as it has the ability to measure significant bearing
faults in motor systems Next temperature sensors specifically thin film thermocouples
(TFTC) and MEMS temperature sensors are discussed These sensors are mounted in the
chamber of a lubricating fluid or in close proximity to a moving component and monitor
temperature fluctuations caused by increased wear Finally acoustic emission (AE)
sensors are discussed which have the ability to detect ultrasonic emissions generated by
plastic deformation in the crystalline lattice of a mechanical component
211 MOTOR CURRENT MONITORING
Monitoring of stator current is used as a method of determining bearing related faults
that produce radial motion between rotor and stator This motion varies the air gap flux
density producing stator currents at predictable frequencies fbng that represent the
bearing faults [10]
bng e vf f m f= plusmn sdot (2)
where fe is the electrical supply frequency m=123hellip and fv is one of the characteristics
frequencies listed in the equations of (1)The bearing damage must be significant to
produce noticeable current fluctuations and it may be difficult to distinguish from current
fluctuations due to voltage imbalances machine saturation or statorarmature faults [2]
7
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
8
212 TEMPERATURE SENSORS
Temperature sensors are largely used in high temperature high load and high speed
environments typically turbine engines and industrial tools Thermocouple sensors
consist of two dissimilar metals joined together at two junctions and when exposed to a
temperature difference generate a thermoelectric voltage dependant only on the material
properties and junction temperature difference [11] In turbine engines these sensors can
be placed on bearing casings or in the lubricant to monitor temperature fluctuations due
to increased friction The thermocouples are fabricated using bulk materials or micro-
machined thin films (TFTC) with the latter providing a faster response small package
and substantially lower production costs [12] The TFTC can be embedded for in situ
monitoring of processes in hostile environments One study successfully embedded
TFTC into electroplated nickel and attained device characteristics on par with surface
mounted devices [13] More recently MEMS temperature sensors have been developed
that consist of micro-machined semiconductor material that undergoes structural
deformation with temperature changes A common implementation is a bent beam
structure whose temperature induced deflection δ is expressed as [14]
( ) 3
2 2
sin12 cos sinz
EA T LEI AEL
α βδ 2β β
∆=
+
(3)
Temperature sensors are well suited for localized fault monitoring and detection but can
only characterize a mechanical system consisting of multiple components through the use
of multiple sensors
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
9
213 ACOUSTIC EMISSION SENSORS
Acoustic emission (AE) sensors have been used to characterize wear in machine
tools and monitor bearing and gear problems in centrifugal pumps [1516] First
developed as a Non-Destructive Testing (NDT) technique to detect cracks in civil
structures these sensors detect acoustic emissions generated by the release of vibration
waves in a crystalline lattice due to plastic deformation or crack growth [16]
Measurements are made using piezoelectric transducers with high natural frequencies
100 kHz to 1 MHz to capture the ultrasonic AE emissions An AE sensor is useful as it
has the ability to detect subsurface cracks in gear teeth or bearings before appearing on
the surface causing further damage [17] More recently MEMS acoustic sensors have
been developed and one design includes multiple transducers on a single substrate
which each detect acoustic emission energy at different frequencies [18] This helps
distinguish spurious acoustic emissions arising from impact and friction from those
arising from plastic deformation When compared to typical piezoelectric sensors the
MEMS devices have lower sensitivities and fail to detect some acoustic emissions
[1819] In addition the acoustic emission signal suffers severe attenuation as is crosses
various interfaces such as a gearbox or bearing casings In one experiment consisting of
a pinion gear and an associated bearing a 44dB attenuation was seen between an AE
sensor placed directly on the pinion to one placed on the bearing casing and in some
cases intermediate loss of the signal was observed [20]
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
10
22 MEMS VIBRATION SENSORS
This thesis focuses on the development of a MEMS vibration sensor for fault
detection as the vibration signature of a mechanical system has the potential to give an
overview of the entire system Many types of MEMS vibration sensors exist with
different operating principals based on resonance the piezoelectric effect and
displacement variation
221 RESONANT SENSORS
Resonant sensors are based on the resonant frequency shift of a fixed-fixed micro-
beam that is excited into resonance using electrostatic thermal or piezoelectric methods
Under an applied axial load there is a shift in the resonant frequency which is expressed
as [21]
2 2
2
1 12 1
nn n
EI Fl2
fl A Eα γ
π ρ= +
I
(4)
where l is the beam length E is Youngrsquos Modulus I is the moment of inertia ρ is the
density A is the cross section area F is the applied axial force αn and γn are mode
dependant coefficients and are 4730 and 0295 for mode 1 in the case discussed The
axial force F is generated by a proof mass attached to one beam end which moves in
response to an external acceleration The performance of resonant sensors are dependant
on the quality factor Q which is the ratio of total energy stored in the system to the
energy lost per cycle and determines the sharpness and amplitude of the resonance peak
A high Q resonator develops a peak that is easily distinguishable in a phase locked loop
control circuit and allows for improved performance and resolution [3] A resonator Q
value is lowered by energy losses to the surrounding fluid attached supports and
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
11
material Losses to the fluid surrounding the MEMS structure is the dominant loss
mechanism for MEMS sensors operated in air and can be greatly reduced by vacuum
packaging which can be accomplished using glass-silicon anodic bonding [22] In
addition to increasing fabrication cost the long term stability of the vacuum degrades
with ultimate device failure stemming from leakage through micro cracks and defects
and out-gassing [22] Support losses arise from the restoring forces they generate and
can be reduce by balanced designs such as a double ended tuning fork (DETF) or
operation in higher modes [3] In addition the sensitivity of a resonant sensor can be
increased by the addition of micro-levers to increase the axial force In one such design a
two stage micro-lever increased the sensitivity by an order of magnitude over a more
conventional single lever arm [2324]
222 PIEZOELECTRIC SENSORS
Piezoelectric materials produce a charge in response to an applied force This is an
intrinsic effect in materials such as Zinc Oxide (ZnO) whereas materials such as Barium
Titanate or Lead Zirconate Titanate (PZT) need to be poled by placing the material in a
strong electric field at an elevated temperature [3] The piezoelectric effect is expressed
mathematical by [25]
i ij j ik kD d T Eε= + (5)
where D represents the electrical displacement (Cm2) d is the charge coefficient (CN)
T is the applied stress matrix (Nm2) ε is the permittivity matrix (Fm) and E is the
electric field matrix (Vm) The subscripts i=123 j=123456 and k=123 represent
the direction to which the physical property is related For micro-sensors poled in the out
of plane direction only the D3 term is relevant and if fabricated using ZnO or PZT only
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
12
the charge coefficients d31 d32 and d33 are relevant [25] The piezoelectric material is
commonly deposited as a thin film on the surface of compliant structures in bulk micro-
machined sensors or incorporated as a layer in surface micro-machining on a similar
structure [26] However the inclusion of multiple materials introduces large temperature
variations causing out of plane bending especially in surface micro-machined devices
[2627] Poled piezoelectric materials could be depolarized which serves to reduce the
piezoelectric affect in the material This can occur if the sensor is exposed to a strong
electric field of the opposite polarity high temperatures in excess of the Curie point or
high mechanical stress Also the piezoelectric effect reduces over time an effect that can
be reduced with the addition of composite elements [28] Foundry processes such as
PolyMUMPS and SoiMUMPS do not incorporate piezoelectric layers thus sensors using
this method of transduction must use a customized process The piezoelectric sensor
outlined in [26] uses polysilicon surface micro-machining similar to PolyMUMPS
however it incorporates a ZnO layer that is not available in the foundry process
223 DISPLACEMENT VARIATION SENSORS
Displacement variation sensors consist of a proof mass connected to a compliant
micro-flexure The movement of the proof mass is sensed using optical electron
tunneling or more commonly capacitive methods A recent optical accelerometer
utilized a novel multilayer nano-grating for in-plane displacement sensing of the proof
mass [29] For small space variations between the nano-gratings a large change in the
optical reflection amplitude of the grating was observed [30] Even though optical
sensors offer resolutions approaching the Brownian noise limit they are not available in
small packages that are easily placed in space constricted areas for a low cost [29]
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
13
The electron tunneling effect is observed when the proximity between two electrodes
one flexible and the other fixed is on the order of 10Ǻ [31] If the two electrodes are
biased with voltage VB a tunneling current It flows between the gap [31]
exp( )t B I tgI V xαinfin sdot minus Φ (6)
where α1 is a constant and equal to 1025 Ǻ-1 eV-frac12 Ф is the effective height of the
tunneling barrier and xtg is the tunneling gap When the flexible electrode moves in
response to acceleration a feedback controller maintains the original distance while
determining the applied acceleration Electron tunneling sensors are able to sense
accelerations in the microg-ng range due to the exponential relation between tunneling current
and gap However these highly sensitive devices are susceptible to thermo-mechanical
noise which could be reduced by operation at low pressure [32] It is also important for
the two electrodes to be coated with metal which is difficult to achieve with foundry
fabrication
Capacitive sensors are based on the parallel plate capacitor and are implemented
using inter-digitated comb fingers in an in-plane lateral or transverse drive or in an out of
plane parallel plate drive [33] These sensors have the advantage of high sensitivity low
power consumption small package low temperature dependence and are easily
integrated with existing foundry fabrication processes [34] The lateral drive offers a fully
linear response but low sensitivity whereas the transverse and parallel plate drives
achieve high sensitivity if gap distances are of a few microns and large capacitance areas
are utilized However both drives have a non-linear response and are prone to pull-in
Also if these sensors are fabricated using a surface micromachining process such as
PolyMUMPS where the ground plane is a few microns above the structural layer they
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
14
are prone to electrical failure due to shorting This was observed in an experiment where
similar sensors were placed in a vibrating environment with a peak of 120g and shorting
between the structural layer and ground plane was found to be the predominate mode of
failure [35] However with certain fabrication processes such as SoiMUMPS this issue
can be neglected as the substrate material below any moving component is removed
Most capacitance sensors are fabricated using an integrated CMOS (Complementary
Metal Oxide Semiconductor)-MEMS technique where the MEMS sensor is fabricated
before during or after the CMOS circuit fabrication [36] Although these sensors offer
high-sensitivity and small parasitic capacitance they suffer from in-plane and out of
plane curling of the beam which reduces the capacitance between adjacent comb fingers
[37] In addition multiple processing steps are required for device fabrication
More than the other transduction methods MEMS capacitive displacement sensors are
attractive for practical implementation in an automobile and therefore is the selected
method for the vibration sensor presented in this work Their low temperature dependence
is an ideal characteristic for an automobile environment that faces varying temperature If
implemented with off-chip readout circuitry they require a standard foundry fabrication
process which provides a strong platform for mass production and low unit cost
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
3 SENSOR STRUCTURE
This chapter presents the physical structure of the three MEMS vibration sensors
which includes the displacement sensing capacitive comb drive and the micro-
flexureproof mass system First the capacitive comb drive and the three variations
implemented the transverse drive lateral drive and differential tri-plate drive are
discussed Next the mechanical structure is presented and includes the micro-flexure
selection quality factor and equations of motion Finally instability that places a limit
on the operating range of the sensor as a result of a gap closing capacitive comb drive is
discussed
31 SENSING
Capacitive sensors are based on the parallel plate capacitor and are commonly
implemented using inter-digitated comb fingers A capacitance can be realized if a voltage
is placed between two closely spaced plates If the fringing electric field is neglected the
capacitance of the parallel plate capacitor Cpp (F) is expressed as [28]
air pppp
k AC
dε
= (7)
where kair=1 is the dielectric constant of air at atmospheric pressure ε=88542times10-12 F m-1
is the permittivity of free space App is the overlapping area (m2) and d is the distance
between the two plates (m)
15
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
16
Figure 3-1 An inter-digitated comb drive where changes in capacitance are generated by either changes in gap distance x0 or in the overlap area y0timest
Figure 3-1
For a set of inter-digitated comb fingers or comb drive shown in the total
capacitance Ctotal is expressed as [28]
0 0
0 0
( ) ( )( ) (Total cy y y yC N t
)x x x xε
+ += + + minus
(8)
where N is the number of comb fingers on the movable comb y0 is the initial comb finger
overlap length (m) tc is the out of plane comb finger thickness (m) and x0 is the inital
finger gap (m) In this work the comb drive is implemented in three arrangements
transverse lateral and differential tri-plate In a transverse comb drive changes in
capacitance are generated by transverse movements along the x-axis resulting in changes
to the gap x0 between the movable and fixed comb fingers The nonlinear relation
between total capacitance CTotalT and displacement is expressed as
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
17
00 0
1 1( ) (Total TC N ty
)x x x xε
= + minus +
(9)
Although the drive sensitivity part is not constant an average sensitivity can be
assumed for small displacements [38]
Total TC xpart
In a lateral comb drive changes in capacitance are generated by lateral movements
along the y-axis resulting in changes to the overlapping area y0timestc The total capacitance
CTotalL between all comb fingers is expressed as
00
2 ( )Total LN tC yx
yε= +
(10)
The sensitivity of the lateral comb drive SL is expressed as
0
2Total LL
C N tSy x
εpart= =
part
(11)
The lateral comb drive was first introduced by W C Tang T H Nguyen and R T
Howe [33] in an electrostatic resonator and has since been used numerous times in both
drive and sense configurations It has the advantage of a fully linear response and
negligible instability but offers low sensitivity The sensitivity for the lateral drive L1
implemented in this work is 0038 pF microm-1
A differential tri-plate drive shown in Figure 3-2 which is a variation of the
transverse drive is implemented to improve sensitivity and reduce noise [39]
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
18
Figure 3-2 An inter-digitated comb drive in a differential tri-plate arrangement where x01 x02 and x03 are the gap distances and x01= x02ltlt x03 to improve sensitivity
In the tri-plate drive the at rest gap distances between the two capacitors formed by any
movable finger is not equivalent For a differential arrangement insuring that x01=
x02ltltx03 the linearity for small displacements (less than 1 microm) is enhanced over the case
of a single ended arrangement where x01=x02=x03=x0 which is equivalent to the
transverse drive This is shown graphically in Figure 3-3 where x01= x02=x03=2 microm in
the lower curve and x01= x02= 2 microm and x03= 15 microm in the upper curve A substantial
improvement in linearity is attained with the tri-plate drive
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
19
0
005
01
015
02
025
03
035
0 02 04 06 08 1
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-3 Comparison of the effect of gap distances on linearity for two case lsquorsquo represents case 1 where the fingers in are implemented in a differential tri-plate arrangement and x01= x02= 2 microm and x03= 15 microm lsquorsquo represents the case 2 where the fingers are implemented in a single ended arrangement and x01= x02=x03=2 microm
The tri-plate drive is also implemented in a differential arrangement where the output
is the change in capacitance rather than the total capacitance resulting from an external
excitation This reduces errors cause by off-axis excitation and common mode noise The
output of the differential tri-plate drive is expressed as [39]
Total Top Total BottomC C C∆ = minus where
003 01
1 1( ) (Total TopC N ty
)x x x xε
= +
minus +
(12)
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
20
002 03
1 1( ) (Total BottomC N ty
)x x x xε
= +
minus +
A graphic comparison of the capacitancedisplacement relation for sensors T1 and T2 is
shown in Figure 3-4 For these gap closing drives it is assumed that for displacements
less that 1 microm the relation is linear indicated by the dashed line above each curve The
sensitivity within this region is approximately 0046 pF microm-1 and 0253 pF microm-1 for T1
and T2 respectively The order of magnitude improvement is attributed to the differential
tri-plate arrangement and thicker comb finger
0
02
04
06
08
1
12
14
16
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
Cap
acita
nce
(pF)
Figure 3-4 lsquorsquo represents the output of the differential tri-plate drive while lsquorsquo represents the output of the transverse drive The dashed line represents the linear approximation for displacements less that 1 microm For the purpose of comparison the output of the transverse drive was equalized to zero at zero displacement
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
21
In this work the transverse drive is first implemented in sensor T1 using the
PolyMUMPS fabrication process Then in an effort to linearize the sensor response the
lateral drive is used in L1 using SoiMUMPS The reduction in drive sensitivity due to the
lateral drive is further reduced by increasing the gap distance to 3 microm which is
recommended by the foundry to ensure a successful device These reductions are
balanced by increasing the number of comb fingers N and selecting a substrate thickness
that is 125 times large than T1 The third device T2 is fabricated using SoiMUMPS and
uses the tri-plate differential drive This design capitalizes on the increase in sensitivity of
a gap closing drive and larger substrate thickness attaining an order of magnitude
increase in sensitivity over T1 and L1 The specifications for each drive along with the
successive improvement in drive sensitivity are outlined in Table 3-1
Table 3-1 Specifications for each capacitive comb drive implemented in this work T1 L1 and T2
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Comb Finger Thickness tc (microm) 2 25 25 Gap Distance x0 (microm) 2 3 2 Number of Fingers N 200 256 30 Overlap length y0 (microm) 45 10 70 Drive Sensitivity (pF microm-1) 0046 0038 0273
32 MECHANICAL ANALYSIS
321 MICRO-FLEXURE SELECTION
The sensor structure consists of a proof mass connected to a compliant micro-flexure
that itself is anchored to the sensor substrate The proof mass displaces in response to an
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
22
applied acceleration and this displacement is sensed using the capacitive comb drive
attached to the proof mass
Proof mass
Folded beam micro-flexure
Anchor
Figure 3-5 Sensor Structure
This section focuses on micro-flexure selection as it should exhibit a large linear
displacement range and have a high lateral stiffness ratio to reduce cross-axis sensitivity
A number of micro-flexures have been used in MEMS research these include the
clamped-clamped crab-leg and folded beam shown in Figure 3-6-through 3-7
The clamped-clamped flexure exhibits a high stiffness ratio however for large lateral
displacements extensional axial forces buildup within the beams resulting in a non-linear
force-displacement relation [40] Within the linear regime the stiffness of the clamped-
clamped flexure is expressed as [40]
3
3b
yEt wk
L=
(13)
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
23
where E is the Youngrsquos Modulus of the beam material (N m-2) tb is the beam out of
plane thickness (m) w is the beam width (m) and L is the beam length (m) The stiffness
ratio is expressed as [40]
2x
y
k Lk w
=
(14)
If L is two orders of magnitude greater than w this results in a ratio of approximately
10000 The linear displacement range is determined using Finite Element Analysis (FEA)
of a clamped-clamped flexure with L=387 microm w=4 microm and tb=2 microm The result of this
simulation is plotted in and compared to a structure with constant stiffness ky
= 035 N m-1 predicted by Eq (13) The linear relation is valid for small deflections that
are less than 2 microm In addition this micro-flexure is susceptible to buckling as a result of
a compressive residual stress that may arise due to the fabrication process [41]
Figure 3-9
In the crab leg design the thigh section is meant to relieve the extensional axial
stress postponing the occurrence of non-linearity until greater lateral displacement [42]
The stiffness of the crab leg flexure is expressed as [43]
3
44
b s s ty
s s t
Et w L LkL L L
κκ
+= +
(15)
where κ=IsIt=(wswt)3 is a ratio of the second moment of inertia of the shin (s) and thigh
(t) ws and wt represent the width and Ls and Lt represent the length of the shin and thigh
regions respectively The stiffness ratio is expressed as [43]
34
4t s s tx
y s t s
w L L Lkk w L L L
αα t
+= +
(16)
The linear displacement range of a crab leg flexure with Ls=285 microm Lt=30 microm ws=wt=3
microm and tb=2 microm is determined using FEA The results are plotted in Figure 3-9 and
compared to a structure of constant stiffness ky=035 N m-1 predicted by Eq (15) The
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
24
non-linearity is greatly reduced when compared to the clamp-clamped flexure with Eq
accurately predicting the deflection until 4 microm However the flexure is more
susceptible to off-axis deflection as there is an order of magnitude reduction in the
stiffness ratio as compared to the clamped-clamped flexure
(15)
The folded beam flexure stands out as it is designed to exhibit a large linear
displacement range and to relieve residual stress due to the manufacturing process by
allowing the beams to expand and contract along the axial direction [40] It was first used
in an electrostatic resonator and has subsequently been used for electrostatic actuation in
a magnetic hard drives and micro-positioning [3344] If the truss section is rigid
ItrussgtgtIbeam then the stiffness of the flexure is expressed as [4043]
3
32b
yEt wk
L=
(17)
The stiffness ratio is expressed as [40]
2x
y
k Lk w
=
(18)
Figure 3-6 Clamped-clamped flexure
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
25
Figure 3-7 Crab leg flexure
Figure 3-8 Folded beam flexure
The linear range of motion of a folded beam flexure with L=304 microm w=4 microm and tb=2
microm is determined using FEA The results are plotted in Figure 3-9 and compared to a
structure with constant stiffness ky=035 N m-1 predicted by Eq (17) The folded flexure
exhibits a response that closely follows the linear prediction through a displacement that
exceeds the previous two flexures while maintaining the stiffness ratio of the clamped-
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
26
clamped flexure Due to the linear response high stiffness ratio and compensation for
residual stress the folded flexure is implemented in each of the three vibrations sensors
presented in this work
Ideal
Clamped- Clamped
Folded Beam
Crab Leg
00
10
20
30
40
50
60
70
80
00 20 40 60 80 100Applied Force (microN)
Dis
plac
emen
t (microm
)
Figure 3-9 Comparison of the non-linear deflection among the three micro-flexures considered clamped-clamped crab-leg and folded beam All flexures were designed to have the same stiffness of 035 N m-1
322 EQUATIONS OF MOTION
The vibration sensor in this work uses four folded beam flexures in parallel with a
total stiffness kT (N m-1)
3
3
24 bT
Et wk kL
= = (19)
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
27
The sensor can be approximated as a lumped mass system undergoing damped harmonic
oscillation When attached to a vibrating structure the system undergoes base excitation
shown in Figure 3-10 and the equation describing the absolute motion of the mass is
[45]
( ) ( )eff M M B M Bm y c y y k y y+ minus + minusampamp amp amp 0= (20)
where meff is the effective proof mass (kg) yM is the absolute proof mass displacement
(m) yB is the base displacement (m) and c is the coefficient of damping (N s m-1)
Figure 3-10 Lumped mass approximation with base excitation yB(t)
Of specific interest is the relative motion of the mass zy (m) with respect to the base
zy=yM - yB
If a sinusoidal base excitation is applied yB
yB(t)=YB sin(ωt)
where YB is the base excitation amplitude (m) ω is the base excitation frequency (rad s-1)
and t represents time the equation governing the relative motion of the proof mass is
expressed as [45]
2 22 sy n y n y Bz z z Y in( )tζω ω ω ω+ + =ampamp amp (21)
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
28
where ζ=05b(km)12 is the damping ratio ωn=(km)12 is the proof massflexure natural
frequency (rad s-1) and ω is the base excitation frequency (rad s-1) Eq (21) shows that
the system can be treated as a case of direct mass excitation shown in Figure 3-11 where
the acceleration applied to the mass a (m s-2) is
a=ω2YB sin(ωt)
Figure 3-11 Lumped mass approximation with direct mass excitation z(t)
By evaluating the Laplace transform of the left side of Eq (21) and making the
substitution s=jω the magnitude of the system transfer function is expressed as
2 2 2
1( )( ) (2n n
H jω2)ω ω ζω ω
=minus +
(22)
In the low frequency range ωltltωn the system response is well approximated from the
DC response
H(ω=0)=1ωn2
For a low frequency external acceleration of peak amplitude A (g) the peak response Zy
(m) is expressed as
2
981y
n
AZωsdot
= (23)
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
29
Therefore by decreasing the system natural frequency the response to a low frequency
excitation can be increased The decrease in ωn can be accomplished by decreasing the
flexure stiffness or increasing the proof mass The stiffness can be decreased by
increasing the flexure length however depending on the fabrication process a maximum
length is placed on beams due to out of plane bending Increasing the proof mass can be
accomplished by increasing the proof mass area However limitations are placed on this
approach which are dependent on the fabrication process For example for overall device
mechanical stability in SoiMUMPS a restriction is placed on the amount of removed
bulk material which is required to free moving structures This places a restriction on the
maximum area of the proof mass
A FEA of the three proposed sensors are done to accurately determine the natural
frequency and flexure stiffness of the structures Table 3-2 outlines the sensor
dimensions and mechanical properties The stiffness and natural frequency were found to
be 14 N m-1 and 3500 Hz 378 N m-1 and 1240 Hz and 433 N m-1and 750 Hz for T1
L1 and T2 respectively and represent a good match with the analytical formula
The effectiveness of the proof massflexure mechanical system can be quantified by
considering the mechanical sensitivity SM (m g-1) is given by
2
981yM
n
VS
A ωpart
= =part
(24)
SM is evaluated to be 002 microm g-1 0161 microm g-1 and 0442 microm g-1 for T1 L1 and T2
respectively Sensors L1 and T2 attain an order of magnitude increase over T1 which is
attributed to the thicker substrate used in SoiMUMPS The nearly three fold increase
from L1 to T2 is a result of a substantially larger proof mass area in T2
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
30
T1 L1 T2 Fabrication Process PolyMUMPS SoiMUMPS SoiMUMPS Beam Length L (microm) 650 1048 1000 Beam Width w (microm) 4 8 8 Beam Thickness tB (microm) 2 25 25 Youngrsquos Modulus E (GPa) 158 169 169 Natural Frequency ωn (Hz) 3500 1240 750 Flexure Stiffness k (N m-1) 14 378 433 Effective Mass meff (kg) 29e-9 56e-8 195e-7 Mechanical Sensitivity (microm g-1) 002 0161 0442
Table 3-2 Dimensions and mechanical properties for the three MEMS vibration sensors outlined in this work
323 QUALITY FACTOR
The quality factor Q is the ratio of total energy stored in the system to the energy
lost per cycle The total quality factor Qtotal is expressed as [3]
s
1 1 1
total f mQ Q Q Q= + +
1 (25)
where 1Qf represents the energy lost to the surrounding fluid 1Qs is the energy lost
through the supports and 1Qm is the energy lost through the sensor material For MEMS
sensors Qf is the dominant component and is the focus of the discussion A prediction of
the quality factor is important as it influences the mechanical noise of the sensor
For MEMS devices operated in ambient air energy is lost to the surrounding fluid
through air damping The quality factor as a function of air damping is expressed as
T
n total
kQcω
= (26)
where ctotal is the total damping coefficient (N s m-1) The effects of air damping are
approximated by two mechanism slide film and squeeze film damping Slide film
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
31
damping is caused by the fluid in the gap between two parallel moving plates If a linear
velocity profile is assumed the coefficient of damping cslide-film is approximated by
air slide filmslide film
Acd
micro minusminus =
(27)
where microair=18e-5 N s m-2 is the absolute viscosity of air at 20degC Aslide-film is the effective
area between the two plates (m) and d is the distance between the two plates (m)
Squeeze film damping arises when the gap between two parallel plates decrease
squeezing the fluid between the two plates creating an opposing force Squeeze film
damping csqueeze-film can be approximated using the following relation [3846]
3
3air plate plate
squeeze film
L Wc
dmicro α
minus = (28)
where Lplate is the effective plate length Wplate is the effective plate width and α is a
parameter dependant on the WplateLplate ratio which can be determined from reference
[46] For the sensor T1 slide film damping is assumed between the proof mass and
substrate while squeeze film damping is assumed between the fixed and movable fingers
on the transverse comb drive The total damping coefficient ctotal is expressed as the sum
of the individual contributions
total slide film squeeze filmc c cminus minus= + (29)
For sensor L1 only slide film damping is assumed between the fixed and movable
fingers on the lateral comb drive For sensor T2 only squeeze film damping is assumed
between the fixed and movable fingers on the tri-plate differential comb drive
summarizes the damping constant and approximate quality factors for each sensor
Table 3-3
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
32
Sensor ctotal (N s m-1) Q T1 617e-7 103 L1 468e-7 631 T2 171e-5 53
Table 3-3 Summary of the theoretical coefficients of damping and quality factors for the three sensors presented in this work
33 DRIVE STABILITY
Gap closing capacitive drives such as T1 and T2 suffer from pull-in instability that
places a limit on drive displacement Pull-in occurs since capacitive readout electronics
apply a voltage driving signal across the fixed and moving comb drive which creates an
electrostatic force that brings the fingers together On a lateral drive this force is
balanced at rest and after a displacement however in gap closing drives the force is only
balanced at rest The voltage signal is time varying with peak VE and is typically a square
wave with a 50 percent duty ratio or sinusoidal and has a frequency a few orders of
magnitude higher than the proof massflexure natural frequency [47] For sensors T1 and
T2 the total electrostatic force Fx-Elect applied to the proof mass due to the capacitive
drives is expressed as
2 1
12
Drivex Elect
CF Vx
part=
part
(30)
where V1=VEradic2 for a sinusoidal voltage signal x is the displacement direction
CDrive=CTotalT for T1 and CDrive=CTotalTop+ CTotalBottom for T2 The electrostatic force
applied on T1 and T2 is expressed
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
33
20
1 2 20 0
1 12 ( ) ( )x Elect T
N ty VFx x x x
εminus
= minus minus +
20
2 2 2 203 01 02 03
1 1 1 12 ( ) ( ) ( ) ( )x Elect T
N ty VFx x x x x x x x
εminus
= minus + minus
minus + minus + 2
(31)
Neglecting the effect of damping the equation governing the motion of the proof mass is
expressed as
eff x Electm a kx F= minus (32)
To determine the pull-in acceleration and displacement Eq (32) was solved in [47] by
plotting the right hand side (RHS) for various x values and then plotting the left hand
side (LHS) as a horizontal line for various applied accelerations The RHS of Eq (32) is
plotted in for T1 and T2 and it is observed that there are two points of
intersection between any horizontal line and the RHS curve The first point of
intersection represents the stable solution while the second the unstable The maximum
point of the RHS curve represents the maximum acceleration that can be applied for a
stable response This is approximately 42g for R1 and 125g for T2 beyond these values
the undesired pull-in effect occurs
Figure 3-12
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
34
000
050
100
150
200
250
300
350
400
450
0 02 04 06 08 1 12 14 16 18
Drive Displacement (microm)
RH
S (micro
N)
Figure 3-12 Graph representing stability regions for gap closing capacitive drives lsquorsquo represents the RHS of Eq (32) for T2 while lsquorsquo represents the RHS of T1
34 ELECTROSTATIC ACTUATION
The sensor T1 is fabricated with an electrostatic lateral comb drive which is used to
actuate the proof mass and allow for verification of the sensor frequency response The
electrostatic force generated when a voltage is applied across a parallel plate capacitor is
expressed as [28]
212
Total Lact act
CF V
ypart
=part
(33)
where Fact is the actuation force in the y-direction CTotalL is the capacitance of the lateral
comb drive expressed by Eq (10) and Vact is the applied actuation voltage across the
capacitor plates (v) For the lateral comb drive shown in Figure 3-1 the force in the y
direction is expressed as
2
0act act
N tF Vxε
= (34)
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
35
A sinusoidal excitation voltage with frequency ωv results in an applied force with a DC
component and a time varying components with frequency 2ωv
35 SECTION SUMMARY
Table 3-4
Table 3-4 Summary of the characteristics of the sensors outlined in this work
summarizes the important characteristics of the three sensors presented in
this work Changes are made in each sensor that progressively improves the mechanical
and capacitive drive sensitivity Since T1 and T2 use gap closing capacitive drives they
are subjected to excitation limitations to maintain stability In addition T2 offers a more
linearized response due to the tri-plate arrangement
T1 L1 T2 Mechanical Sensitivity (microm g-1) 002 0161 0442 Capacitive Drive Sensitivity (pF microm-1) 0046 0038 0253 Maximum Excitation (g) 42 na 12g Maximum Excitation for Linearized Response (g)
39 na 115
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
4 CAPACITANCE TO VOLTAGE READOUT CIRCUIT
Capacitance measurements can be made by applying an AC excitation signal on one
node of a capacitor and detecting the flow of charge using a charge amplifier as shown
in Figure 4-1 The amplitude of the output voltage VCA (V) is determined by the feedback
capacitor Cf and the amplitude of the excitation signal VE and is expressed as [48]
sensorCA E
f
CV VC
= sdot (35)
Figure 4-1 Ideal charge amplifier
This basic charge amplifier has been used as a building block for capacitance to
voltage readout circuitry for many capacitance based sensors In an off-chip
implementation it has been used as a readout for an open-loop accelerometer in [49] and
for a MEMS based power disconnect in [38] In the circuit utilized in this work shown in
R1 is required to keep the proof mass of the MEMS sensor at a defined
potential R2 prevents the output from drifting and saturating the amplifier and C1 blocks
any DC component in the sinusoid at the negative input of op-amp 1 [49] If R1 Rf and C1
are selected to be large ie 5 MΩ 2 MΩ and 1 nF respectively the output of the
charge amplifier can be closely approximated by Eq (35) In this work the impedances
Figure 4-2
36
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
37
at the negative input of op-amp 1 were balanced at the positive input by Cb and Rb to
reduce the voltage offset caused by the amplifierrsquos input bias current and to improve the
noise performance [5051]
Figure 4-2 Capacitance to voltage readout circuit which is made up of four
elements the charge amplifier voltage amplifier demodulator and low pass filter
Overall the circuit consists of four elements the charge amplifier voltage amplifier
demodulator and a RC low pass filter The output of the charge amplifier is followed by
a variable gain voltage amplifier with output VVA
VVA= VCAR3R2
This voltage is then directed to the demodulator with output VDM expressed as
VDM=VVAVE10
The output of the demodulator is then low pass filtered with cut-off frequency ωc of
1591 Hz to yield an approximate DC signal The output of the demodulator is a positive
AC signal with a DC offset with the low pass filter effectively removing the AC
component The output of the low pass filter VLP is approximately
VLP = VDM 2
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
38
The DC signal VLP represents the rest capacitance of the sensor as determined by the
comb drive
The circuit was simulated using Orcad and the signals at various nodes are shown in
and The output of the readout circuitry at the low pass filter is
closely approximated by
Figure 4-3
Figure 4-3 Orcad simulation results of the capacitance to voltage readout circuit
The excitation voltage VE is 75 VPK and the output of the voltage amplifier VVA is
approximately 13125 VPK
Figure 4-4
23
220E
LP sensorf
V RV CR C
= sdot (36)
The sensitivity of the readout circuit SRO (V pF-1) is expressed as
23
220ELP
ROsensor f
V RVSC Rpart
= =part C
(37)
With the components shown in Figure 4-2 SRO is approximately 141 V pF-1
15V VVAVE 10V
0
-10V
Time
400us -15V
430us 435us 405us 410us 415us 425us 440us415us
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
39
Figure 4-4 Orcad simulation results showing the sinusoidal output of the
demodulator VDM with a 984 VPK-PK amplitude and the DC output of the low pass
filter VLP at approximately 5 V
Time
400us 410us 415us 425us 430us 435us 440us
0
-15V
VDM 15V 10V VLP
-10V
405us 420us
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
5 NOISE ANALYSIS
51 MECHANICAL-THERMAL NOISE
For small signal sensors the micro-structures are susceptible to mechanical noise
arising from collisions of the surrounding fluid molecules [52] This noise applies a
fluctuating force on the proof mass with spectral density [52]
4N BF k= Tc (38)
where FN is the noise force (NradicHz) kB=138times10-23 JK is Boltzmannrsquos constant and T is
the absolute temperature (K) The mechanical noise displacement is then expressed as
[4552]
22 2
4
1
B totalN
Tn n
k TcX
f fkf f
= minus + Q
(39)
where XN is the noise displacement (mradicHz) kT is the flexure spring constant f is the
excitation frequency (Hz) and fn is the system natural frequency (Hz) If f is much less
than fn then using Eq (23) the noise equivalent acceleration an (gradicHz) is well
approximated by
41981
B nn
k TamQ
ω=
(40)
Therefore to reduce the mechanical noise the sensor should have a large proof mass This
can be achieved using a large area and a fabrication process with a thick structural layer
A high quality factor is also required however MEMS sensors operated in air suffer from
high damping which greatly reduces Q which could be overcome by vacuum packaging
40
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
41
The noise equivalent accelerations for the sensors outlined in this work are 35 microg Hz-frac12
019 microg Hz-frac12 and 027 microg Hz-frac12 for T1 L1 and T2 respectively The order of magnitude
decrease in L1 and T2 over T1 is due to the increase in proof mass size and the
corresponding decrease in the natural frequency
52 ELECTRICAL NOISE
In active devices the primary sources of noise are voltage and current noise which
appear as flicker noise (1f) at low frequency and white noise high frequency while
thermal noise exists in the resistors as white noise [51] To measure the small variations
in capacitance of the MEMS device the noise sources are included in the analysis A
noise analysis is done in PSpice using the AD712 operational amplifier model provided
by Analog Devices The model adds a voltage source in series with the positive input to
include the effect of the input noise voltage and a current source between each input
(positive and negative) and ground to include the effect of the input noise currents As
well Orcad includes the thermal noise of resistors in the analysis At the node VVA of the
circuit shown in Figure 4-2 the noise voltage spectral density eno is shown in
The total output noise voltage (RMS) at the output is the square root of the area
under the noise voltage power spectrum eno2 and is shown in Figure 5-2 Since the
circuit has a low pass filter with a cutoff frequency of 1591 Hz the total voltage noise
Eno is approximately 30 microVrms
Figure
5-1
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
42
Figure 5-1 Orcad simulation of the noise voltage spectral density eno at the output of the voltage amplifier
Figure 5-2 Total output noise Eno at the voltage amplifier which is expressed as Eno=( int eno
2)12
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
6 FABRICATION OVERVIEW
This section briefly overviews the two foundry fabrication processes that are used for
the sensors developed in this work [5354] The first is PolyMUMPS which is a three
layer polysilicon surface micromachining process The second is SoiMUMPS which is a
bulk micromachining process that uses silicon as the structural layer
61 POLYMUMPS
In this work only the first two polysilicon layers POLY0 and POLY1 are used in the
T1 sensor layout Fabrication of the MEMS sensor begins on top of a silicon wafer that is
doped with phosphorus to prevent charge feed-through and coated with a layer of silicon
nitride for electrical isolation shown in Figure 6-1 A The first polysilicon layer POLY0
is deposited at a thickness of 05 microm and is subsequently patterned using lithography
and etched using reactive ion etching (RIE) as shown in Figure 6-1 B POLY0 is a non-
mechanical layer and in this work is used as a ground plane Next a phosphosilicate
glass (PSG) layer which acts as a sacrificial layer is deposited using low pressure
chemical vapor deposition (LPCVD) then patterned using lithography and etched RIE
as shown in Figure 6-1 C The PSG allows for the creation of suspended structures and
its thickness of 2 microm sets the distance between the suspended structures and the ground
plane In areas where the PSG is etched away the suspended structures are anchored
Next the first structural polysilicon layer POLY1 is deposited at a thickness of 2 microm
patterned and etched as shown in Figure 6-1 D POLY1 represents the mechanical layout
of the sensor as it defines the flexures proof mass and the fixed and movable comb
fingers
43
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
44
Figure 6-1 PolyMUMPS fabrication process for a simple cantilever beam
62 SOIMUMPS
The MEMS sensor is fabricated using a silicon-on-insulator (SOI) wafer that consists
of four layers the silicon structural layer insulating oxide substrate and bottom-side
oxide as shown in Figure 6-2 A To define the sensor structures the silicon layer is
patterned using lithography and etched using deep reactive ion etching (DRIE) as shown
in Figure 6-2 B Next the bottom side oxide substrate and insulating oxide are patterned
using lithography and etched using RIE DRIE and wet etching respectively as shown
in Figure 6-2 C This step is required over any area where the structural layer is required
to be movable In this work the substrate needs to be removed under the flexures proof
mass and movable comb fingers The silicon structural layer is available in thicknesses
of 10 microm or 25 microm
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
45
Figure 6-2 SoiMUMPS fabrication process for a simple cantilever beam
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
7 EXPERIMENTAL RESULTS
71 MEMS SENSOR ndash T1
The sensor T1 is fabricated first to explore device sensitivity response to excitation
and possible design errors The 45times45mm sensor die is attached and wire bonded to a
68PGA socket to facilitate testing First the sensor is mounted on a six degree of freedom
robotic manipulator with an attached 01 microm probe tip to evaluate the capacitance-
displacement relation Shielded coaxial cables are soldered to the two pins of the variable
capacitor and connected to a capacitance to digital readout board (Analog Devices AD
7746) Then using the probe tip the proof mass is moved in increments of approximately
02 microm Figure 7-1 shows the experimental data points plotted along with the theoretical
capacitancedisplacement curve according to Eq (9) The experimental data points follow
the theoretical curve well however for displacements greater that 12 microm instability
resulting in greater deviation from the theoretical prediction is observed This is
attributed to the pull-in effect which limits the displacement range of the transverse
comb drive For displacements less than 1 microm an average drive sensitivity of 0024 pF
microm-1 is observed and is indicated by a dashed line and the overall device sensitivity is
000048 pf g-1 This represents a decrease of 54 from the theoretical value of 000092
pF g-1 This discrepancy is attributed to parasitic capacitance which affects the sensor
response in two ways it adds a DC offset and it minimizes the capacitive drive
sensitivity The parasitic capacitance arises between the sensor and substrate the sensing
lines on the 68PGA package bonding pads and substrate and any wiring in the readout
circuit For example large bonding pads on the order 100micromtimes100microm (the size used in
46
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
47
T1) which are required to simplify the wire bonding process can add about one pico-
farad of capacitance to the sensor [55] This degrades the sensitivity of the device which
is in the femto-farad range Parasitic capacitance can be greatly reduced by packaging the
sensor and readout together eliminating the need of long coaxial cables and other wiring
or ideally by fabricating both on the same substrate which eliminates the requirement of
bonding pads
0
01
02
03
04
05
06
07
08
0 02 04 06 08 1 12 14 16 18 2
Displacement (microm)
Capa
cita
nce
(pF)
Figure 7-1 The solid line represents the theoretical capacitance versus displacement relation while the solid points represent the experimental values The dashed line represents the assumed linear relation for displacements less than 1microm The experimental average drive sensitivity is approximately 0024 pF microm-1 For displacements greater than 12 microm drive instability is observed and is attributed to the pull-in effect
Next to evaluate the mechanical sensitivity the electrostatic comb drive is used to excite
the flexureproof mass system into resonance to determine the sensor natural frequency
The sensor was mounted inside a custom fabricated vacuum chamber to reduce the effect
of air damping on the response [56] The frequency of the sinusoidal voltage signal
excitation signal is varied in increments of 200 Hz and the amplitude and phase of the
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
48
output signal is shown in Figure 7-2 The resonance peak is located at approximately
3600 Hz and is accompanied with a phase shift of 168deg The experimental mechanical
sensitivity as determined by Eq (24) is 0019 microm g-1
0
05
1
15
2
25
3
35
4
45
400 800 1200160020002400280032003600400044004800520056006000
Frequency (Hz)
Am
plitu
de (V
)
-80
-60
-40
-20
0
20
40
60
80
100
120
Phas
e (deg
)
Figure 7-2 lsquotimesrsquo represents the amplitude and solid line represents the phase response The resonance peak is located at 3600 Hz and accompanied by a phase shift of 168deg
Next the sensor is mounted near the free end of a clamped-free aluminum beam with
a fundamental natural frequency of approximately 23 Hz By displacing the beam free
end and releasing it accelerations in the range of 0-20 g are generated The accelerations
are verified using a commercial accelerometer Model 352C67 manufactured by PCB
Piezotronics At rest the sensor produces a RMS capacitance of 0014 pF which
represents the noise in the system When divided by the experimental device sensitivity
of 000046 pf g-1 this corresponds to a noise equivalent acceleration of 30g Therefore
for applied accelerations less than 30g the sensor produces no observable response as the
minute displacements are buried in the noise signal For an applied acceleration of 20 g
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
49
the theoretical prediction estimated a displacement of 040 microm resulting in a capacitance
increase of 10 fF The lack of a distinguishable response is due to the low capacitive
drive and mechanical sensitivity which both can be attributed to the minimal 2microm
structural layer thickness The mechanical sensitivity could be increased by enlarging the
proof mass area however since the out of plane stiffness is of the same order of
magnitude as the primary direction this may lead to shorting between the structural layer
and the ground plane which are only 075 microm apart
The limitations of the sensor are found to be dependant on the fabrication process
and the integration of the readout circuit and sensor The primary fabrication limitation is
the 2 microm structural layer thickness of PolyMUMPs which impedes the capacitive drive
and mechanical sensitivities In addition the excessive wiring required to connect the
AD7746 readout board to the sensor contributes to a large parasitic capacitance that
further reduces the device sensitivity The fabrication limitation can be reduced by
selecting the SoiMUMPS process which provides a structural layer thickness of 25 microm
In addition the parasitic capacitance can be reduced by using a custom capacitance to
readout circuit mounted close to the 68PGA socket eliminating the AD7746 board and
long wiring
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
50
Figure 7-3 A SEM micro-graph showing the sensor TI fabricated using PolyMUMPS The four folded beam flexures had a length of 650 microm and total stiffness of 14 Nm The effective proof mass was 29e-9 kg and the system had a natural frequency of 3500 Hz
Figure 7-4 A SEM micro-graph of the transverse comb drive implemented in T1 with 200 comb fingers a gap distance of 2 microm overlap length of 45 microm and thickness of 2 microm The average experimental sensitivity was found to be 0024 pF microm-1 for displacements of less than 1 microm
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
51
72 READOUT CIRCUITRY
The readout circuitry is fabricated using discrete components on a prototype board
The charge amplifier and voltage amplifier are mounted as close as possible to the
68PGA socket to minimize electronic noise and parasitic capacitance Due to space
constraints the multiplier and low pass filter are placed on a separate prototype board In
addition to the circuit components of 01 microF and 1 microF capacitors are placed in
parallel to the power supplies of each IC component to minimize any high frequency
noise
Figure 4-2
An input excitation signal VE of 75 VPK at a frequency of 100 kHz is applied to the
circuit and all IC components were powered with plusmn15 VDC supply voltages The readout
circuit is first calibrated using discrete capacitors from 1-4 pF and the sensitivity is found
to be 13 V pF-1 compared to the theoretical sensitivity of 141 V pF-1 The calibration
curve is shown in Figure 7-5 The discrepancy in the output voltage is attributed to the
capacitors which could differ by plusmn025 pF from their ideal values The MEMS sensor is
then connected to the readout circuit and an output voltage of 49 V was observed which
corresponds to a total capacitance of 35 pF
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
52
0
1
2
3
4
5
6
0 05 1 15 2 25 3 35 4
Capacitance (pF)
Out
put V
olta
ge (V
)
Figure 7-5 The solid line represents the theoretical capacitance to voltage relation expressed by Eq (37) with a sensitivity of 141 V pF-1 The solid points represent the experimental data points when discrete capacitors 1-4 pF are used in the readout circuit The horizontal dashed line represents the output voltage of the MEMS sensor at rest and corresponds to a capacitance of 35 pF
73 MEMS SENSORndash L1
The sensor L1 is fabricated using the SoiMUMPS process and tested with the analog
capacitance to voltage readout circuit First with the MEMS vibration sensor at rest the
output of the readout circuitry at various stages is measured using an Agilent 54600
oscilloscope The top trace in represents the input excitation signal from a BK-
Precision 4011a Function Generator while the bottom trace represents the output of the
voltage amplifier The output of the demodulator and the low pass RC filter is shown in
The results exhibit a slight variation from the theoretical values which is
attributed to non-ideal capacitors and resistors
Figure 7-6
Figure 7-7
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
53
Figure 7-6 Oscilloscope trace of the input excitation voltage VE asymp 75 VPK and output of the voltage amplifier VVA asymp 103 VPK Readings were taken with the MEMS sensor at rest
Figure 7-7 Oscilloscope trace of the demodulator output VDM asymp 76 VPK-PK and output of the low pass filter VLP asymp 36 V Readings were taken with the MEMS sensor at rest
Next the sensor and readout circuit are mounted on the six degree of freedom robotic
manipulator to determine the capacitive drive sensitivity The proof mass is displaced in
increments of 02 microm and the output voltage of the readout circuitry at the low pass filter
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
54
VLP for each step is recorded Figure 7-8 displays the capacitance values referenced to
zero which are determined using the experimental readout circuit sensitivity value
0
005
01
015
02
025
03
035
04
0 1 2 3 4 5 6 7 8 9 10Displacement (microm)
Cap
acita
nce
(pF)
Figure 7-8 lsquorsquo represents the experimental data points with the dashed line representing the fitted curve Solid line represents the theoretical data points according to Eq (10)
The experimental sensitivity for the lateral drive is found to be 0025 pF microm-1 and
represents a 32 decrease from the theoretical value of 0038 pF microm-1 The discrepancy
is attributed to parasitic capacitance that is inherent to the sensor and packaging
Next the dynamic response of the sensor is tested by mounting it near the free end of
a clamped-free aluminum beam The output of the readout circuitry at the low pass filter
is connected to an Agilent 35670 Dynamic Signal Analyzer The experimental setup is
shown in Figure 7-9 and Figure 7-10 To capture only the AC change in the output
voltage the signal analyzer is set to AC coupling to remove the DC offset
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
55
Function Generator Power
Supplies
Readout (DM amp LP)
Oscilloscope
Dynamic Signal Analyzer Clamped-
Free Beam MEMS Sensor and Readout (CA amp VA)
Figure 7-9 Experiment setup which includes the MEMS senor and readout board mounted on the clamped-free beam and test equipment
MEMS Sensor
68 PGA Package
Commercial Accelerometer
Figure 7-10 Close-up of the MEMS sensor and readout board mounted on the test beam along with the commercial piezoelectric accelerometer Only the charge amplifier (CA) and the voltage amplifier (VA) are mounted on the same prototype board as the MEMS sensor due to space constraints
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
56
To calibrate the MEMS sensor the results are compared with the true beam
acceleration which is determined by the commercial piezoelectric accelerometer
mounted near the beam free end as shown in Figure 7-10 This accelerometer is factory
calibrated and exhibits a sensitivity of 01 V g-1 and resolution of 016 mg As the beam
is set into vibration the response of each sensor is shown in Figure 7-11 The dynamic
response of the MEMS sensor matches the piezoelectric accelerometer exhibiting the
expected exponentially decreasing sinusoidal beam vibration
Figure 7-11 Figure comparing the response of a commercial piezoelectric accelerometer and MEMS sensor (L1) to a beam vibration A) Dynamic response of the piezoelectric accelerometer representing the true beam vibration with a peak voltage 114 V corresponding to an applied acceleration of 114 g B) Dynamic response of the MEMS sensor with a peak voltage of 006 V
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
57
Next as the beam is set into vibration two traces are exported from the signal
analyzer and are shown in The top trace represents the time domain signal of
the beam vibration while the bottom represents the frequency spectrum The maximum
voltage VLPMAX of the time domain signal is 73 mV which corresponds to acceleration
of approximately 17 g The frequency spectrum shows a peak at approximately 23Hz
which corresponds to the fundamental natural frequency of the beam
Figure 7-12
Figure 7-12 Sensor dynamic response to the vibration of the clamped-free beam and represents a direct export from the dynamic signal analyzer (Agilent 35670) Top trace is the time domain signal of the low pass filter output where VLPMAXasymp73mV which represents a maximum acceleration of approximately 15g The bottom trace represents the frequency spectrum with a peak at 23Hz which corresponds to the beam natural frequency
The maximum output voltage VLPMAX is measured for six accelerations tests
conducted within a range of 585-18 g The linear response of the sensor and readout
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
58
circuitry is shown graphically in Figure 7-13 The theoretical sensitivity S (mV g-1) is
found to be 862 mV g-1 and is expressed as
RO L MS S S S= times times (41)
The experimental sensitivity for the sensor which represents the slope of the linear curve
fitted to the experimental data is 4258 mV g-1 and represents a 50 difference from the
theoretical value This is largely attributed to the presence of parasitic capacitance which
can be further reduced by incorporating the MEMS device and circuit components on the
same substrate using a CMOS fabrication process The decrease in sensitivity is also
attributed to manufacturing irregularities that result in variations to the ideal mechanical
characteristics such as beam dimensions and Youngrsquos Modulus The discernable signal
from L1 can be attributed to its mechanical sensitivity which represents an order of
magnitude increase over T1
0
001
002
003
004
005
006
007
008
009
01
0 5 10 15 20
Input Acce leration (g)
Out
put V
olta
ge (V
)
Figure 7-13 Experimental results showing the response of the sensor to various accelerations in the range of 0-20 g where lsquorsquo represents the experimental data points and the solid line represents the fitted curve The experimental sensitivity was found to be 4258 mV g-1
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
59
The minimum detectable acceleration (g) is determined by the Brownian noise of the
MEMS sensor and the electronic noise of the readout circuitry along with the device
sensitivity and is found by analyzing the time domain signal with the sensor at rest
Min Det Acceleration = RMS Noise Voltage Device Sensitivity
This resulted in a RMS voltage of 1658 mV and represents an equivalent acceleration of
038g This value sets the detection limit for the sensor and only accelerations above this
value can be detected The noise voltage is two orders of magnitude higher than the
predicted value of 003 mV and the increase is attributed to 60 Hz electronic noise from
the surrounding test environment This can be reduced by completely shielding the
MEMS sensor and readout circuitry and noise from the function generator and power
supplies By comparison in a bandwidth of 1591 Hz a CMOS-MEMS sensor developed
in [57] exhibited a minimum detectable acceleration of 004 g an order of magnitude
lower than L1 In the same bandwidth an Analog Devices accelerometer ADXL103
fabricated using their proprietary Integrated Microelectromechanical Systems (iMEMS)
process that combines the sensor and signal conditioning circuitry achieves a minimum
detectable acceleration of 00044 g two orders of magnitude lower that L1
74 SECTION SUMMARY
This section overviewed the experimental results for two MEMS sensors that are
fabricated and tested The first sensor T1 did not achieve a distinguishable response
which is attributed to the low mechanical sensitivity and the influence of parasitic
capacitance brought on by the MEMS device packaging and wiring The second sensor
L1 is fabricated in the SoiMUMPS process and achieved good results as the thicker
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
60
structural layer improved the capacitive drive and mechanical sensitivities in turn
increasing the overall device sensitivity summarizes the experimental results
for each sensor
Table 7-1
Table 7-1 Summary of experimental results
T1 L1 Fabrication PolyMUMPS SoiMUMPS Capacitive Drive Sensitivity (pF microm-1) 0024 0025 Device Sensitivity (g mV-1) na 4258 Noise Floor (g Hz-frac12) na 00098 Min Detectable Acceleration (g) 30 038
The sensor noise floor which determines the minimum detectable acceleration is
determined by the Brownian noise of the MEMS sensor and the electronic noise of the
readout circuit The latter is the dominant component and represents the limit for off-
chip readout circuitry For the noise floor to reach the physical limit of Brownian noise
the MEMS sensor and readout circuitry must be incorporated on the same substrate using
an integrated MEMS-CMOS process and be well shielded
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
8 CONCLUSION AND FUTURE WORK
81 CONCLUSION
This work presented three MEMS capacitive sensors for the detection of low
frequency vibrations The first sensor T1 used the PolyMUMPS foundry fabrication
process while the latter two L1 and T2 used SoiMUMPS Experimental results are
presented for T1 and L1 and the experiments conducted include static tests on a six
degree of freedom robotic manipulator with a 01 microm probe tip and dynamic tests on a
vibrating cantilever beam Capacitance measurements are initially made using a
commercial capacitance to digital readout board (AD 7746) In an attempt to reduce
parasitic capacitance and environmental electronic noise due to the wiring required to
connect the MEMS package to the AD 7746 an analog capacitance to digital readout
board is fabricated using discrete components
Experimental results for T1 produced a transverse comb drive sensitivity of 0024 pF
microm-1 that is 54 less than the predicted value The discrepancy could be due to fringe
capacitances that are neglected in the theoretical calculations parasitic capacitance that
arise between the sensor and substrate sensing lines on the packaging and wiring and
any out of plane deflection cause by residual stress that results in a misalignment between
the comb drive fingers In addition the sensor produced no observable response when
subjected to vibration in the range of 0-20g as the minute displacements are buried in
noise This is due to the low sensitivity of the device which is limited by the substrate
thickness of 2 microm in PolyMUMPS Increasing the thickness effects the achievable natural
frequency and comb drive capacitance and results in an increase in overall device
61
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
62
sensitivity This is addressed by selecting the SoiMUMPS process which has a structural
layer thickness of 25 microm and the potential to increase device sensitivity by an order of
magnitude The sensor L1 is also designed with a lateral comb drive that has an intrinsic
linear response In addition the fabricated sensor is tested with a custom made
capacitance to voltage readout circuit The experiments resulted in a capacitance drive
sensitivity of 0025 pF microm-1 that is 32 lower than the predicted value which implies
that parasitic capacitance is still an important issue The sensor exhibited a sensitivity of
4258 mV g-1 when subjected to cantilever beam vibrations between 0-20g and a
minimum detectable acceleration of 038g
The third MEMS sensors T2 is designed and submitted for fabrication using the
SoiMUMPS process It increases sensitivity by two orders of magnitude over L1 to an
estimated 170 mV g-1 using a tri-plate drive and is implemented in a differential
arrangement to reduce noise However ultimately the device performance is limited by
parasitic capacitance that reduces device sensitivity and electronic noise that increase the
minimum detectable acceleration
82 FUTURE WORK
Future work should investigate the use of a foundry CMOS process to implement the
MEMS sensor and readout electronics This can be done in two ways
bull MEMS sensor and readout circuit are fabrication using two different
processes For example SoiMUMPs for MEMS and a 08 micron CMOS
process for the readout which are then combined using wafer level
packaging
bull MEMS sensor and readout circuit are fabricated on the same substrate
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
63
In addition future work should also focus on developing a more complete FEA model
that includes a couple-field analysis which investigates the combined effects of the
mechanical electrostatic and capacitive components of the MEMS sensor This allows
for a more accurate prediction of the sensor response as well as the intrinsic sensor
parasitic capacitance
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
64
References
[1 V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[1] V Wowk Machinery Vibration Measurement and Analysis McGraw-Hill Inc 1991
[2] S Nandi H A Toliyat and Xiaodong Li Condition monitoring and fault diagnosis
of electrical motors-a review IEEE Transactions on Energy Conversion vol 20 pp 719-729 2005
[3] S P Beeby G Ensel and M Kraft MEMS Mechanical Sensors Artech House
2004 [4] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998 [5] W J Fleming Overview of automotive sensors IEEE Sensors Journal vol 1 pp
296-308 2001 [6] D S Eddy and D R Sparks Application of MEMS technology in automotive
sensors and actuators Proceedings of the IEEE vol 86 pp 1747-1755 1998 [7] G Brasseur Robust automotive sensors Instrumentation and Measurement
Technology Conference1997 IMTC97 Proceedings of the IEEE Sensing Processing Networking vol 2 pp 1278-1283 vol2 1997
[8] D R Sparjs Packaging of microsystems for harsh environments IEEE
Instrumentation amp Measurement Magazine vol 4 pp 30-33 2001 [9] R G Azevedo Jingchun Zhang D G Jones D R Myers A V Jog B Jamshidi
M B J Wijesundara R Maboudian and A P Pisano Silicon carbide coated MEMS strain sensor for harsh environment applications IEEE 20th International Conference on Micro Electro Mechanical Systems pp 643-646 2007
[10] R R Schoen T G Habetler F Kamran and R G Bartfield Motor bearing
damage detection using stator current monitoring IEEE Transactions on Industry Applications vol 31 pp 1274-1279 1995
[11] R Komanduri and Z B Hou A review of the experimental techniques for the
measurement of heat and temperatures generated in some manufacturing processes and tribology Tribology International vol 34 pp 653-682 10 2001
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
65
[12] H D Bhatt R Vedula S B Desu and G C Fralick Thin film TiCTaC thermocouples Thin Solid Films vol 342 pp 214-220 326 1999
[13] A Datta H Choi and X Li Batch Fabrication and Characterization of Micro-Thin-
Film Thermocouples Embedded in Metal Journal of The Electrochemical Society vol 153 pp H89-H93 2006
[14] B Ando S Baglio N Pitrone N Savalli and C Trigona Bent Beam MEMS
Temperature Sensors for Contactless Measurements in Harsh Environments Instrumentation and Measurement Technology Conference Proceedings IMTC 2008 pp 1930-1934 2008
[15] T J Holroyd Machine condition and dynamic diagnostics via acoustic emission
IEEE Seminar on On-Line Monitoring Techniques for the Off-Shore Industry pp 51-54 1999
[16] S Dolinsek and J Kopac Acoustic emission signals for tool wear identification
Wear vol 225-229 pp 295-303 4 1999 [17] N Tandon and A Choudhury A review of vibration and acoustic measurement
methods for the detection of defects in rolling element bearings Tribology International vol 32 pp 469-480 8 1999
[18] D Ozevin S P Pessiki D W Greve and I J Oppenheim A MEMS transducer for
detection of acoustic emission events IEEE Sensors pp 4 pp 2005 [19] I J Oppenheim D W Greve D Ozevin D R Hay T R Hay S P Pessiki and N
L Tyson Structural tests using a MEMS acoustic emission sensor Smart Structures and Materials 2006 Sensors and Smart Structures Technologies for Civil Mechanical and Aerospace Systems Proceedings of the SPIE vol 6174 pp 26-35 2006
[20] T Toutountzakis C K Tan and D Mba Application of acoustic emission to
seeded gear fault detection NDT amp E International vol 38 pp 27-36 1 2005 [21] V Ferrari A Ghisla D Marioli and A Taroni Silicon resonant accelerometer
with electronic compensation of input-output cross-talk Sensors and Actuators A Physical vol 123-124 pp 258-266 923 2005
[22] S Choa Reliability of MEMS packaging vacuum maintenance and packaging
induced stress Microsystem Technologies vol 11 pp 1187-1196 1030 2005
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
66
[23] S X P Su H S Yang and A M Agogino A resonant accelerometer with two-stage microleverage mechanisms fabricated by SOI-MEMS technology IEEE Sensors Journal vol 5 pp 1214-1223 2005
[24] A A Seshia M Palaniapan T A Roessig R T Howe R W Gooch T R
Schimert and S Montague A vacuum packaged surface micromachined resonant accelerometerJournal of Microelectromechanical Systems vol 11 pp 784-793 2002
[25] S Kon and R Horowitz A High-Resolution MEMS Piezoelectric Strain Sensor for
Structural Vibration Detection IEEE Sensors Journal vol 8 pp 2027-2035 2008 [26] D L DeVoe and A P Pisano Surface micromachined piezoelectric accelerometers
(PiXLs) Journal of Microelectromechanical Systems vol 10 pp 180-186 2001 [27] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Micromachined inertial sensors Proceedings of the IEEE vol 86 pp 1640-1659 1998
[28] C Liu Foundation of MEMS Pearson Prentice Hall 2006 [29] U Krishnamoorthy R H Olsson III G R Bogart M S Baker D W Carr T P
Swiler and P J Clews In-plane MEMS-based nano-g accelerometer with sub-wavelength optical resonant sensor Sensors and Actuators A Physical vol 145-146 pp 283-290 0 2008
[30] B E N Keeler D W Carr J P Sullivan T A Friedmann and J R Wendt
Experimental demonstration of a laterally deformable optical nanoelectromechanicalsystem grating transducer Optics Letters vol 29 pp 1182-1184 0601 2004
[31] CH Liu and T W Kenny A high-precision wide-bandwidth micromachined
tunneling accelerometer Journal of Microelectromechanical Systems vol 10 pp 425-433 2001
[32] T K Bhattacharyya A Ghosh and D Paul Physical modelling of a MEMS based
electron tunneling accelerometer IEEE SAS pp 101-106 2008 [33] W C Tang T H Nguyen and R T Howe Laterally Driven Polysilicon Resonant
Microstructures Sensors and Actuators vol 20 pp 25-32 1115 1989 [34] N Yazdi N Yazdi F Ayazi and K Najafi Micromachined inertial sensors
Proceedings of the IEEE vol 86 pp 1640-1659 1998
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
67
[35] D M Tanner J A Walraven K S Helgesen L W Irwin D L Gregory J R Stake and N F Smith MEMS reliability in a vibration environment 38th Annual International Reliability Physics Symposium Proceedings of IEEE pp 139-145 2000
[36] H Baltes O Brand A Hierlemann D Lange and C Hagleitner CMOS MEMS -
present and future The Fifteenth IEEE International Conference on Micro Electro Mechanical Systems pp 459-466 2002
[37] H Luo G Zhang L R Carley and G K Fedder A post-CMOS micromachined
lateral accelerometer Journal of Microelectromechanical Systems vol 11 pp 188-195 2002
[38] H K Chu J K Mills and W L Cleghorn MEMS-based power disconnect for 42-
V automotive power systems Journal of MicroNanolithography MEMS and MOEMS vol 7 pp 013010 2008
[39] Y Sun S N Fry D P Potasek D J Bell and B J Nelson Characterizing fruit fly
flight behavior using a microforce sensor with a new comb-drive configuration Journal of Microelectromechanical Systems vol 14 pp 4-11 2005
[40] R Legtenberg A W Groeneveld and M Elwenspoek Comb-drive actuators for
large displacements Journal Micromechanics and Microengineering vol 6 pp 320-329 1996
[41] Horsley D A Horsley D A Horowitz R amp Pisano A P (1998)
Microfabricated electrostatic actuators for hard disk drives IEEEASME Transactions on Mechatronics 3(3) 175-183
[42] A P Pisano Resonant-structure micromotors Micro Electro Mechanical Systems
1989 Proceedings an Investigation of Micro Structures Sensors Actuators Machines and Robots IEEE pp 44-48 1989
[43] G K Fedder Simulation of microelectromechanical systems PhD dissertation
University of California Berkeley 1994 [44] J Dong D Mukhopadhyay and P M Ferreira Design fabrication and testing of a
silicon-on-insulator (SOI) MEMS parallel kinematics XY stage Journal of Micromechanics and Microengineering vol 17 pp 1154-1161 2007
[45] S S Rao Mechanical Vibrations 4th ed Prentice Hall 2005 [46] R P van Kampen M J Vellekoop P M Sarro and R F Wolffenbuttel
Application of electrostatic feedback to critical damping of an integrated silicon
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000
68
capacitive accelerometer Sensors and Actuators A Physical vol 43 pp 100-106 5 1994
[47] M Bao H Yang H Yin and S Shen Effects of electrostatic forces generated by
the driving signal on capacitive sensing devices Sensors and Actuators A Physical vol 84 pp 213-219 91 2000
[48] H Baltes O Brand G K Fedder C Hierold J G Korvink O Tabata Enabling
Technologies for MEMS and Nanodevices Wiley-VCH 2004 [49] M Kraft Closed loop digital accelerometer employing oversampling conversion
PhD dissertation Coventry University 1997 [50] M Napoli C Olroyd B Bamieh and K Turner A novel sensing scheme for the
displacement of electrostatically actuated microcantilevers Proceedings of the American Control Conference pp 2475-2480 vol 4 2005
[51] S Franco Design with operational amplifiers and analog integrated circuits 3rd
ed Mcgraw Hill 2002 [52] T B Gabrielson Mechanical-thermal noise in micromachined acoustic and
vibration sensors IEEE Transactions on Electron Devices vol 40 pp 903-909 1993
[53] POLYMUMPS Design Handbook Revision 110 2005 [54] SOIMUMPS Design Handbook Revision 40 2004 [55] M E Motamedi MOEMS SPIE Press 2005 [56] P Hassanpour Design and Analysis of Microelectromechanical Resonant
Structures PhD dissertation University of Toronto 2008 [57] H Luo G K Fedder and L R Carley A 1 mG lateral CMOS-MEMS
accelerometer The Thirteenth Annual International Conference on Micro Electro Mechanical Systems pp 502-507 2000