MEMS and power harvesting Embedding sensors in large-scale built

infrastructure

Ashwin A. Seshia

aas41@cam.ac.uk

Cambridge University Nanoscience Centre

Department of Engineering

Microfabricated sensors

• Integration with Electronics

• System complexity

• Increased sensitivity

• Capability for arrays

• Low power

• Integration with energy harvesting

• Small size

• Capability for multi-domain integration

• Portable

• Mobile access

• Batch manufacturability

• Low cost

Sensing requirements Embedding sensors in large-scale built infrastructure

• Accuracy of data must meet model requirements.

(configure sensing modalities and accuracies based on

model demands).

• Data must be available on-demand (e.g. the potential for

sensors to be integrated into wireless networks).

• Ease of integration with decision system (e.g. sensors for

performance-based maintenance of built infrastructure).

• Ease of deployment and maintenance: low-cost, low-power

and minimal maintenance. (integrate sensing nodes with

power harvesting devices).

• Strain sensitive

• = f()

• Resolution < 1n

• Power < 1 mW

Optimum Mode for the DETF

Double-ended Tuning Fork Device (DETF)

Electrode

DETF

Electrode

Vibration direction

Frequency

Ga

in

Resonance

= 0

> 0

Tine displacement

Sensing principles –

Micromachined Resonant Sensors

Resonator sensor interface

oscillator output

Energy input

dissipated

energy

dissipated

energy

Transducer

physical/chemical stimulus

-140

-130

-120

-110

-100

-90

-80

-70

-60

10.753 10.7535 10.754 10.7545 10.755

Frequency [MHz]

Sp

ec

tra

l D

en

sit

y [

dB

c/H

z]

F-/A-/D

conversion

ASIC power dissipation

The oscillator core is designed to sink about 200mA current @3.3V supply.

Motional impedances under 15 kohm to meet GSM noise specification.

B. Bahreyni et al, 2007

Allan Deviation

0.1

1

10

100

1000

0.1 1 10 100 1000

INTEGRATION TIME (s)

ALLA

N D

EV

IA

TIO

N (

PP

B) TCXO (Agilent 33220A)

MEMS SE mode

Lee et al, EDL, 2008

Long-Term Stability

Rob Candler, 2005

STRAIN SENSORS

MEMS Resonant Strain Gauges

• Multi-axis sensing of strain

• Temperature compensation

• Capped under vacuum

• Polysilicon surface micromachining

Strain sensor response

Wall crack

Steel strip

Wall anchors PCB (wireless

unit/sensor

interface) Silicon chip

Uniaxial strain sensors

• Strain resolution: 23 p in 1 Hz

• Scale factor: 150 Hz/m

Collaboration with CNR Italy

FIELD DEPLOYMENT

MEMS-based extensometer

225 µm

M. Ferri et al, 2011

Temperature compensation

M. Ferri et al, 2010

Deployment in Prague Metro

Measurements in Prague Metro

Collaborators: CNR Italy and Czech Technical University

TILT SENSORS

Micro-Electro-Mechanical

Resonant Tilt Sensor

DETF_2

DETF_1

q

Tilt Axis

Z

Freq.1

Increase

Freq.2

Decrease

F≈Mproofg

*sinθ

Tensile

Stress

Compress

Stress

X. Zou et al, 2012

Micro-Electro-Mechanical

Resonant Tilt Sensor

X. Zou et al, 2012

ENVIRONMENTAL IMMUNITY

Introduction – Vibration mode

localization • Consider an array of identical vibratory masses coupled through

weak springs –

• Irregularities inhibit propagation of vibration causing vibration

localization.

• Extent of localization depends on

– magnitude of induced disorder

– Strength of coupling spring with weaker springs leading to

greater localization.

Device 1 Device 2 Weak coupling Perfectly

symmetric When

perturbed

L+LL

M

LL

M M MInitially m1=m2; k1=k2

P. Thiruvenkatanathan et al, 2010.

Common mode rejection

• An important sensor metric – minimizing sensitivity of the sensor

to unwanted variables (for e.g..: environmental drift).

• Various compensation techniques are usually required in most

MEMS devices.

• Mode localization provides outputs that are nearly 1000 times

more sensitive to the measurand but are also relatively

insensitive to environmental drift.

• Since eigenstates are deduced from the relative amplitudes of

vibration of both nearly identical devices any effects on stiffness

and / or mass due to envtal drifts affect both simultaneously.

Device 1 Device 2 Weak coupling )2,1(;

40

i

k

k

u

u

ci

i

)2,1(;40

i

m

k

k

m

u

u

ci

i

P. Thiruvenkatanathan et al, 2010.

Intrinsic common mode rejection

The variations in the eigenstates are studied as a function of temperature and pressure. These variations are expected to be independent to first order.

Nearly one order of magnitude more insensitive to temperature variations.

300 310 320 330 3402.0

2.5

3.0

3.5

4.0

4.5

5.0 Relative shift in first eigenstate

Relative shift in second eigenstate

Theoretical

Rela

tive v

ariations (

%)

Temperature (K)(a)

10 100 10001.0

1.5

2.0

2.5

3.0

3.5

4.0

Eig

ensta

te v

ariation (

%)

Pressure (mTorr)

Variation in first eigenstate

Variation in second eigenstate

Theoretical

(a)

P. Thiruvenkatanathan et al, IEEE Transactions on Ultrasonics, Ferroelectrics and Frequency Control, 2010.

Example: charge sensing

After calibrating and measuring initial response of the two coupled resonators, the variation in the first eigenstate were studied for varying charge input and compared with corresponding resonant frequency variations:

CHARGE INPUT

Sense port 1

Sense port 2

270.54 270.57 270.600.0

1.0m

2.0m

3.0m

4.0m

5.0m

6.0m

Tra

nsm

issio

n (

lin

ea

r s

ca

le)

Frequency (kHz)

0 fC

113 fC

188 fC

263 fC

338 fC

413 fC

489 fC

(a)

Measure responses from resonators 1 (a) and 2 (b) for varying charge input

SEM image of mode-localized electrometer

270.54 270.57 270.600.0

2.0m

4.0m

6.0m

8.0m

10.0m

Tra

nsm

issio

n (

lin

ea

r s

ca

le)

Frequency (kHz)

0 fC

113 fC

188 fC

263 fC

338 fC

413 fC

489 fC

(b)

9.0m

9.2m

9.4m

9.6m

P. Thiruvenkatanathan et al, IEEE IFCS 2010.

VIBRATION ENERGY

HARVESTING

Energy harvesting for ultra-low

power sensors

• Environmental sensors operating on scavenged energy.

• Sensor packaged for operation in harsh environments.

• Sensors embedded in low power distributed sensor networks for environmental monitoring.

• Energy harvesting from ambient mechanical, fluidic and thermal sources.

• Application context: Natural and built environments.

Wall crack

Steel strip

Wall anchors PCB (wireless

unit/sensor

interface)

Silicon chip

Uniaxial strain sensors

0 500 1000 1500 2000 25000

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1x 10

-6

frequency

am

plitu

de

Real-world applications

• Intermittent, irregular and broadband nature of real vibrations.

• Arrayed linear, MDOF or non-linear approaches for vibration energy

harvesting must be considered.

• Increased device complexity for non-linear mechanisms.

Y. Jia et al, submitted to Smart Materials and Structures, 2012 (under review).

Energy Harvesting from

ambient vibration

Z. J. Wong et al, PowerMEMS 2009.

Vibration Energy Harvesting

• Aims o Converting ambient vibration to useful energy

o Self sustain low power wireless or remote systems

• Challenges o Limited power levels from conventional directly forced resonance

o Confined frequency response despite broadband nature of real vibration

Vibrational

excitation Vibrational

excitation

)(3 tFxkxxcxm m mx +cx + k (t )x +mx 3 = F (t )

Direct resonance Parametric resonance

Advantages of parametrically

excited systems

• Stores an order more energy in the system:

significantly improved mechanical-to-electrical

transduction efficiency.

• Offers non-linear resonant peaks: this widens

frequency band.

• Demonstrated:

– 10x improvement in harvested power densities.

– 3x improvement in the bandwidth for a given

order of resonance.

Numerical Models

Y. Jia et al, Energy Harvesting Network Dissemination Event, 2012

Energy harvesting comparison

Y. Jia et al, Energy Harvesting Network Dissemination Event, 2012

Measured harvested energy

Y. Jia et al, submitted to Smart Materials and Structures, 2012 (under review).

MEMS parametric harvester

Y. Jia et al, PowerMEMS 2012 (to be presented).

MEMS parametric harvester

Y. Jia et al, PowerMEMS 2012 (to be presented).

Summary

• MEMS meets many certain key requirements

for the monitoring of large-scale built

infrastructure.

• Ongoing research is addressing aspects

relating to improved accuracy, environmental

immunity, packaging/interfacing, reliability,

power dissipation and integration with energy

harvesting systems.

• Much work still remains to be done to validate

MEMS technology in real-world infrastructure

monitoring applications.

Acknowledgments

• Collaborators

– Kenichi Soga, Robert Mair, Cam

Middleton, Jim Woodhouse, Mark

Welland, Cambridge University

– Alberto Roncaglia, Matteo Ferri,

CNR-IMM, Italy

• Funding

– Engineering and Physical Sciences

Research Council

– Technology Strategy Board

– British Council

– European Science Foundation

– US Army Soldier Systems Center

– Royal Society

• Students and Post-doctoral

researchers

– Jize Yan

– Jia Yu

– Pradyumna Thiruvenkatanathan

– Xudong Zou

– Joshua Lee

– Andreja Erbes

– James Ransley

– Gary Choy

– Zi Jing Wong

– Behraad Bahreyni

– Luca Belsito