Semiconductor Detectors and Electronics · Solid State Detectors and Electronics – Introduction Helmuth Spieler TRIUMF Summer Institute 2007 LBNL 3 When first introduced in the

TRIUMF Summer Institute

July 6 – 20, 2007

Semiconductor Detectors and Electronics

Helmuth SpielerPhysics Division

Lawrence Berkeley National LaboratoryBerkeley, CA 94720

[email protected]

These course notes and additional tutorials athttp://www-physics.lbl.gov/~spieler

or simply Google “spieler detectors”

More detailed discussions inH. Spieler: Semiconductor Detector Systems, Oxford University Press, 2005

Solid State Detectors and Electronics – Introduction Helmuth SpielerTRIUMF Summer Institute 2007 LBNL

2

Outline

Detector Applications and RequirementsEnergy MeasurementsFast TimingPosition Sensing

Strip and Pixel DetectorsSignals and Noise

Sensor Physics ISignal Magnitude and Fluctuations

Fano FactorSignal Formation

Induced CurrentSensors + Amplifiers

Semiconductor Diodespn-JunctionsDepletion Width

Sensor Physics IICharge Collection in

pad detectors

Charge collection in strip and pixeldetectors

TrappingSensor Materials

ElectronicsElectronic NoisePulse ShapingThreshold Discriminator SystemsTiming MeasurementsDigital Electronics

Analog-to-Digital ConversionPulse Processing

Readout Systems

Detector Structures II – Pixel DevicesHybrid pixel devices, CCDs, SiliconDrift Chambers, DEPFETs, MAPS

Why Things Don’t Work


3

When first introduced in the 1960s,semiconductor detectors wereused primarily for high-resolutionx-ray and gamma spectroscopy.

Typical signal-to-noise ratio ~1000

0 500 1000 1500 2000ENERGY (keV)

10

10

10

10

10

10

6

5

4

3

2

CO

UN

TS

NaI(Tl) SCINTILLATOR

Ge DETECTOR

(J.Cl. Philippot, IEEE Trans. Nucl. Sci. NS-17/3 (1970) 446)


4

A major scientific and industrial application is x-ray fluorescence

When excited by radiation of sufficient energy, atoms emit characteristic x-rays thatcan be used to detect trace contaminants.

Experimental Arrangement

The incident radiation can be broad-band, as long as it contains components of higherenergy than the atomic transitions of the atoms to be detected.


5

High-rate high-resolutionx-ray fluorescence systemsopened a wealth of applications inscience and industry.

Example:Detection of trace contaminants

Human blood sample prior tointroduction of unleaded gasoline:

log scale!

Necessary to measure lowintensity peaks adjacent to verystrong signals

concentrations in ppm

Fig. courtesy of Joe Jaklevic, LBNL


6

As explained in previous lectures, the interaction mechanisms of gamma raysdepend on energy. Absorption coefficient µ in Si:

0.001 0.01 0.1 1 10 100 1000PHOTON ENERGY (MeV)

0.001

0.01

0.1

1

10

100

1000

10000

ABSO

RP

TIO

N C

OEF

FIC

IEN

T (c

m-1

)

RAYLEIGH

COMPTON

PHOTOELECTRIC

PAIR

Fraction of particle interactions:0

1 exp( )N

xN

µ= − −


7

Distinguish between

• Detection

• Energy measurement

Detection possible with fraction ofphoton energy.

For full energy absorption, typicallymultiple processes are involved.

If the detector volume is sufficientlylarge, the full photon energy will bemeasured.

“Full energy peak”:

From Knoll,Radiation Detection and Measurement


8

With a small detector volume,secondary particles canleave the detector volume,so only a fraction of theincident photon energyis measured.

From Knoll,Radiation Detection and Measurement


9

Fraction of photons fully absorbed vs. energy for different thicknesses ofSi and Ge.


10

Thin Si detectors have ns collectiontimes and provide ps time resolution.

Nuclear time-of-flight system:

Flight path s= 20 cm, σt ≈ 20 ps

Spieler et al., Z. Physik A278 (1976) 241


11

Position Sensing – Rethinking Requirements

“Traditional” Si detector system Tracking Detector Module (CDF SVX)for charged particle measurements 512 electronics channels on 50 µm pitch


12

Spectroscopy systems highly optimized!By the late 1970s improvements were measured in %.

Separate system components:1. detector2. preamplifier3. amplifier

adjustable gainadjustable shaping

(unipolar + bipolar)adjustable pole-zero cancellationbaseline restorer

Beam times typ. few days with changing configurations, so equipment must bemodular and adaptable.

In large systems as in HEP power dissipation and size are critical, so systemsare not designed for optimum noise, but adequate noise, and circuitry designedfor specific detector requirements.


13

In high energy physics the largest Si detector arrays are used for positionsensing.

The ability to pattern electrodes on micron scales coupled with high rate capability andradiation resistance makes semiconductor detectors the system of choice for at small radii.Robust operation also facilitates large area systems (ATLAS: 60 m2, CMS: 260 m2).

E

PARTICLETRACK


14

Resolution determined by precision of micron scale patterning

Two options:

1. Binary Readout 2. Analog Readout

Interpolationyieldsresolution < pitch -15 -10 -5 5 10 15 [ m]µ

TRANSVERSE DIFFUSION

PARTICLE TRACK

To Threshold Discriminators

Position resolution set directly bystrip pitch, i.e.strip center-to-center distance:

12x

pitchσ =Relies on transverse diffusion: x colltσ ∝e.g. collt ≈10 ns ⇒ σx= 5 µmInterpolation precision depends on S/N and strip pitch


15

In tracking systems the biggest challenge is reducing mass.

Impact parameter resolution

⇒ a) the ratio of outer to inner radius should be large

b) the resolution of the inner layer σ1 sets a lower bound on the overall resolution

c) the acceptable resolution of the outer layer scales with r2 /r1.

If the layers have equal resolution σ1= σ2 = σ

The geometrical impact parameter resolution is determined by the ratio of the outer to inner radius.

The obtainable impact parameter resolution decreases rapidly from

σb /σ = 7.8 at r2 /r1 =1.2 to σb /σ = 2.2 at r2 /r1= 2 and σb /σ < 1.3 at r2 /r1 > 5.

The inner radius is limited by the beam pipe, typically r= 5 cm.At high luminosities, e.g. LHC, radiation damage is a serious concern, which tends to drive the innerlayer to larger radii.

2

12

2

21

2

1/1

/11

−

+

−

≈

rrrrb

σσ

2

12

22

21

12

12

122

12

212

1//1

−

+

−

=

−

+

−

≈rrrrrr

rrr

rb

σσσσσ


16

Amount of material and its distribution is critical:

Small angle scattering:[ ]

0 0

0.0136 /1 0.038 lnrms

GeV c x xp X X⊥

Θ = + ⋅

Assume a Be beam pipe of x= 1 mm thickness and R= 5 cm radius:

The radiation length of Be is X0= 35.3 cm ⇒ x/X0= 2.8.10-3

⇒ For p⊥= 1 GeV/c the scattering angle Θrms= 0.56 mrad.

This corresponds to σb = RΘrms= 28 µm

Exceeds the impact parameter resolution typically achievable by the detector:

For σ = 10 µm and r2 /r1 ≈ 3: σb ≈ 25 µm.

Scattering originating at small radii is more serious ⇒Important to limit material at small radii.

Especially critical at ILC!

For comparison: 300 µm of Si (typ. strip detector) → 0.3% X0


17

Two-Dimensional Position Sensing

Crossed Strips

n readout channels ⇒ n2 resolution elements

y

x


18

Problem: Ambiguities with multiple simultaneous hits (“ghosting”)

HITGHOST

n hits in acceptance field ⇒ n x-coordinatesn y-coordinates

⇒ n2 combinations

of which 2n n− are “ghosts”


19

Reduce ambiguities by small-angle stereo

In collider geometries often advantageous, as z resolution less important than rϕ

The width of the shaded area subject to confusion is 22

1

tanp

L pp

α +

Example: ATLAS SCT uses 40 mrad small-angle stereoTwo single-sided strip detectors glued back-to-back

L

p1

p2

αW


20

Eliminate “ghosting” in non-projective configurations.

Example: hybrid pixel detector (also CCDs, MAPS)

Example: ATLAS pixel detector (just installed), ~1m2, 80 million channelsChallenge: power dissipation, but with optimized design

power per m2 comparable to strip detector systems.

READOUTCHIP

SENSORCHIP

BUMPBONDS

READOUTCONTROLCIRCUITRY

WIRE-BOND PADS FORDATA OUTPUT, POWER,AND CONTROL SIGNALS


21

Baseline Fluctuations (Electronic Noise)

Choose a time when no signal is present.

Amplifier’s quiescent output level (baseline):

In the presence of a signal, noise + signal add.

Signal: Signal+Noise (S/N = 1)

TIME

TIMETIME


22

Measurement of peak amplitude yields signal amplitude + noise fluctuation

The preceding example could imply that the fluctuations tend to increase the measured amplitude, sincethe noise fluctuations vary more rapidly than the signal.

In an optimized system, the time scale of the fluctuation is comparable to the signal peaking time.

Then the measured amplitude fluctuates positive and negative relative to the ideal signal.

Measurements taken at 4 differenttimes:

noiseless signal superimposed forcomparison

S/N = 20

Noise affects

Peak signal

Time distribution

TIME TIME

TIME TIME


23

Electronic noise is purely random.

⇒ amplitude distribution isGaussian

⇒ noise modulates baseline

⇒ baseline fluctuationssuperimposed on signal

⇒ output signal has Gaussiandistribution

Measuring ResolutionInject an input signal with known charge using a pulse generator set to approximate thedetector signal shape.

Measure the pulse height spectrum. peak centroid ⇒ signal magnitudepeak width ⇒ noise (FWHM= 2.35 Qn)

0

0.5

1

Q s /Q n

NO

RM

ALI

ZED

CO

UN

T R

ATE

Q n

FWHM=2.35Q n

0.78


24

Signal-to-Noise Ratio vs. Detector Capacitance

if Ri x (Cdet + Ci) >> collection time,

peak voltage at amplifier input ss sin

det i

i dtQ QV

C C C C= = =

+∫

↑Magnitude of voltage depends on total capacitance at input!

R

AMPLIFIER

Vin

DETECTOR

CC idet i

v

q

t

dq

Qs

c

s

s

t

t

t

dt

VELOCITY OFCHARGE CARRIERS

RATE OF INDUCEDCHARGE ON SENSORELECTRODES

SIGNAL CHARGE


25

The peak amplifier signal SV is inversely proportional to the total capacitance at theinput, i.e. the sum of

1. detector capacitance,

2. input capacitance of the amplifier, and

3. stray capacitances.

Assume an amplifier with a noise voltage nv at the input.

Then the signal-to-noise ratio

1S

n

VSN v C

= ∝

• However, /S N does not become infinite as 0C → (see Part III)

• The result that / 1/S N C∝ generally applies to systems that measure signal charge

Solid State Detectors and Electronics – Sensor Physics I Helmuth SpielerTRIUMF Summer Institute 2007 LBNL

Sensor Physics – Part I

Any form of elementary excitation can be used to detect the radiation signal.

An electrical signal can be formed directly by ionization.

Incident radiation quanta impart sufficient energy to individual atomic electrons to form electron-ionpairs (in gases) or electron-hole pairs (in semiconductors and metals).

Other detection mechanisms areExcitation of optical states (scintillators)Excitation of lattice vibrations (phonons)Breakup of Cooper pairs in superconductorsFormation of superheated droplets in superfluid He

Typical excitation energiesIonization in gases ~30 eVIonization in semiconductors 1 – 5 eVScintillation ~10 eVPhonons meVBreakup of Cooper Pairs meV


Band Structure in Crystals

Example: Lattice structure of diamond, Si, Ge (“diamond lattice”)

dimension a: lattice constant Diamond: 3.56 ÅGe: 5.65 ÅSi: 5.43 Å

a


Extent of wavefunctions of typical constituent atoms:

(following Shockley)

CARBON ( = 6)Z SILICON ( =14)Z GERMANIUM ( = 32)Z

1 AAPPROXIMATE SCALE:


Crystal Bonds

Si

Si

Si

Si

Si

Si

Si

SiSILICON ATOM WITH FOURVALENCE ELECTRONS

SYMBOLIC PLANE VIEW USINGLINES TO REPRESENT BONDS

SILICON “CORES” WITH ELECTRON“CLOUDS” SHOWING VALENCE PAIR BONDS


When isolated atoms are brought together to form a lattice, the discrete atomic states shift to formenergy bands:

Filled band formed by bondingstates: Ψ= Ψa + Ψa

(Ψa = wavefunction of individualatom)

Empty band formed by anti-bonding states: Ψ= Ψa − Ψa

(vanishing occupancy at mid-pointbetween atoms)

Each atom in the lattice contributesits quantum states to each band:The number of quantum states in the band is equal to the number of states from which the band wasformed.The bands are extended states, i.e. the state contributed by an individual atom extends throughout thecrystal.

E

E

p

s

ENERGY

FORBIDDENGAP

ANTI-BONDING STATES(EMPTY ORBITALS) CONDUCTION BAND−

BONDING STATES( FILLED ORBITALS)

VALENCE BAND−

DENSITY

ISOLATED ATOMS

METAL COVALENTSOLID


Energy band structure

Typical band gaps(valence – conduction band)

Ge 0.7 eV

GaAs 1.4 eV

Si 1.1 eV

Diamond 5.5 eV

DISTANCEDENSITY OF STATES

ENE

RG

Y

ENE

RG

Y

CONDUCTION BAND CONDUCTION BAND

VALENCE BAND VALENCE BAND

CORE ELECTRONS

FORBIDDEN GAP

FORBIDDEN GAP



At 0K all electrons occupy bonding states, completely filling the valence band.

If an electric field is applied to the crystal, no current can flow, as this requires that the electrons acquireenergy, which they can’t, as no higher energy states are available in the valence band.

If energy is imparted to a bond by incident radiation,for example a photon, the bond can be broken,

• exciting an electron into the conduction band and

• leaving back a vacant state in the valence band, a“hole”.

SiSi

SiSi

SiSi

SiSi

SiSi

Si

Si

Si

Si

Si

Si

Si

Si

INCIDENT PHOTON BREAKS BOND


The electron can move freely in its extended state.The hole can be filled by an electron from a nearby atom, thereby moving to another position.

The motion of the electron and hole canbe directed by an electric field.

Holes can be treated as positive chargecarriers just like the electrons

However, they tend to move more slowlyas hole transport involves sequentialtransition probabilities (the wavefunctionoverlap of the hole and its replacementelectron).

SiSi

SiSi

SiSi

SiSi

SiSi

Si

Si

Si

Si

Si

Si

Si

Si

NET MOTIONOF ELECTRON

NET MOTIONOF HOLE

MOTION OFREPLACEMENTELECTRONS

ELECTRIC FIELD


Ionization energy in solids is proportional to the band gap

small band gap ⇒ ~ conductorelectric field smallDC current >> signal current

large band gap ⇒ insulatorhigh electric fieldsmall signal charge+ small DC currentexample: diamond

moderate band gap ⇒ semiconductorhigh electric field“large” signal chargesmall DC current, but“pn-junction” required.

examples: Si, Ge, GaAs

Although phonons have been represented as a penalty that increases the ionization energy, asmentioned above they are another form of elementary excitation that can be used to measure the signal.More about this in the final lecture.


Detector Sensitivity

Example: Ionization signal in semiconductordetectors

a) visible light (energies near band gap)

Detection threshold = energy required toproduce an electron-hole pair ≈ band gap

In indirect bandgap semiconductors (Si),additional momentum required:provided by phonons

(from Sze)


Band Structure

Energy of the conduction and valence band edges vs. wave vector (momentum)(from Sze)

Note that in Si and Ge theminimum of the conduction bandis offset from the maximum of thevalence band.

⇒ Promotion of an electronfrom the valence to theconduction band using anenergy equal to the minimumgap spacing requiresadditional momentumtransfer


b) high energy quanta ( gE E )

It is experimentally observed that the energyrequired to form an electron-hole pair exceeds thebandgap.

Why?

When particle deposits energy one mustconserve both

energy and momentum

momentum conservation not fulfilled bytransition across gap

⇒ excite phonons

C.A. Klein, J. Applied Physics 39 (1968) 2029


Phonon energy vs. momentum (wavevector k)

In a semiconductor ionizationdetector ~60% of thedeposited energy goes intophonon excitation.

Instead of detecting electron-hole pairs, detect heat or phonons

Energy scale: 10 meV ⇒ lower energy threshold

Another possibility: Breakup of Cooper pairs in superconductorsThe energy gap 2∆ (order 1 meV) is equivalent to the band gap in semiconductors.


Signal Fluctuations in a Scintillation Detector

Example: Scintillation Detector - a typical NaI(Tl) system(from Derenzo)

Resolution of energy measurement determined bystatistical variance of produced signal quanta.

1E N NE N N N

∆ ∆= = =

Resolution determined by smallest number of quanta inchain, i.e. number of photoelectrons arriving at firstdynode.

In this example

12% rms = 5% FWHM

3000

EE∆

= =

Typically 7 – 8% obtained, due to non-uniformity of lightcollection and gain.

511 keV gamma ray

⇓25000 photons in scintillator

⇓15000 photons at photocathode

⇓3000 photoelectrons at first dynode

⇓3.109 electrons at anode

2 mA peak current


Fluctuations in the Signal Charge: the Fano Factor

The mean ionization energy exceeds the bandgap for two reasons

1. Conservation of momentum requires excitation of lattice vibrations

2. Many modes are available for the energy transfer with an excitation energy lessthan the bandgap.

Two types of collisions are possible:

a) Lattice excitation, i.e. phonon production (with no formation of mobile charge).

b) Ionization, i.e. formation of a mobile charge pair.


Fano Factor in Semiconductors

The incident photon energy goes into two modes:

1. ionization (excitation of electrons into the conduction band)characterized by the bandgap energy gE

2. phonons (lattice vibrations),characterized by the average phonon energy phE

The total energy

0 phph g gE n E n E= + (1)

Since the absorbed energy is fixed, any variation in energy going into phonon modes must be balancedby an opposite change in ionization, so

phph g gn E n Eδ δ= (2)

Thus, provided the number of phonon excitations is large, so that Gaussian statistics apply

( ) ( )2 2

2 2ph ph

g ph phg g

E En n n

E Eδ δ

= ⋅ = ⋅

(3)


Assume that the ratio of the total energy required to form an electron-hole pair to the bandgap /i gE E isconstant. Then for a given number of electron-hole pairs gn the total absorbed energy

0i

g gg

EE n E

E

=

(4)

and eqn 1 can be rewritten as

1g iph g

ph g

E En n

EE

= −

(5)

Inserting this into eqn 3 yields

( )2 1ph ig g

g g

EEn n

E Eδ

= − ⋅

(6)

Simple statistics would imply that ( )2g gn nδ = , so eqn 6 shows that the variance in the number ofelectron-hole pairs is reduced by the Fano factor

1ph i

g g

EEF

E E

= −

(7)

Since / 3i gE E ≈ and ph gE E , the Fano factor 1F < .


In Silicon phE =0.037 eV gE = 1.1 eV iE = 3.6 eV

for which the above expression yields F= 0.08, in reasonable agreement with the measured valueF =0.1.

⇒ The variance of the signal charge is smaller than naively expected:

0.3Q QNσ ≈

A similar treatment can be applied if the degrees of freedom are much more limited and Poissonstatistics are necessary.

However, when applying Poisson statistics to the situation of a fixed energy deposition, which imposesan upper bound on the variance, one can not use the usual expression for the variance var N N=Instead, the variance is 2( )N N F N− = as shown by Fano [1] in the original paper.

An accurate calculation of the Fano factor requires a detailed accounting of the energy dependent crosssections and the density of states of the phonon modes. This is discussed by Alkhazov [2] and vanRoosbroeck [3].

References: 1. U. Fano, Phys. Rev. 72 ( 1947) 262. G.D. Alkhazov et al., NIM 48 (1967) 13. W. van Roosbroeck, Phys. Rev. 139 (1963) A1702


Intrinsic Resolution of Semiconductor Detectors

2.35 2.35 2.35FWHM i Q i ii

EE FN F FEE

Eε ε∆ = ⋅ = ⋅ = ⋅

Si: iE = 3.6 eV F = 0.1

Ge: iE = 2.9 eV F = 0.1

Detectors with good efficiency for thisenergy range have sufficiently smallcapacitance to allow electronic noise of~100 eV FWHM, so the variance of thedetector signal is a significantcontribution.

At energies >100 keV the detector sizesrequired tend to increase the electronicnoise to dominant levels. 0 5 10 15 20 25

ENERGY (keV)

0

50

100

150

200

250

RE

SOLU

TIO

N(e

VFW

HM

)

SILICON

GERMANIUM


3. Signal Formation

Semiconductor Detectors are Ionization Chambers:Detection volume with electric field

Energy deposited → positive and negative charge pairs

Charges move in field → external electrical signal current

If ( )i d iR C C⋅ + collection time ct the peak voltage at the amplifier inputdet

ss

i

QV

C C=

+

R

DETECTOR

CVC iid i

v

q

t

dq

Qs

c

s

s

t

t

t

dt

VELOCITY OFCHARGE CARRIERS

RATE OF INDUCEDCHARGE ON SENSORELECTRODES

SIGNAL CHARGE

AMPLIFIER


Ionization chambers can be made with any medium that allows charge collection to a pair of electrodes.

Medium can be gasliquidsolid

Crude comparison of relevant properties

gas liquid soliddensity low moderate highatomic number Z low moderate moderateionization energy εi moderate moderate lowsignal speed moderate moderate fast

Desirable properties:

• low ionization energy ⇒ 1. increased charge yield dq/dE

2. superior resolution1 1

/i

i

EE

E N E E

∆∝ ∝ ∝

• high field in detection volume ⇒ 1. fast response2. improved charge collection efficiency (reduced trapping)


Formation of a High-Field Region

To form a current that can be measured in the external circuit, the signal charge carriers must bebrought into motion. This is done by establishing a field in the detection volume. Increasing the field willsweep the charge more rapidly from the detection volume.

The conduction band is only empty at 0K.

As the temperature is increased, thermal excitation can promote electrons across the band gap into theconduction band.

Pure Si: carrier concentration ~ 1010 cm-3 at 300K (resistivity ≈ 400 kΩ.cm)

Since the Si lattice comprises 5 . 1022 atoms/cm3, many states are available in the conduction band toallow carrier motion.

In reality, crystal imperfections and minute impurity concentrations limit Si carrier concentrations to~1011 cm-3 at 300K.

This is too high for use in a simple crystal detector.A crystal detector is feasible with diamond,but the charge yield is smaller due to the larger band gap.High-field region with low DC current in semiconductorsis most easily achieved utilizing a pn-junction.

⇒ Introduction of impurities to control conductivity.


Doping

The conductivity of semiconductors can be controlled by introducing special impurities.required concentrations: ~1012 – 1018 cm-3

Replacing a silicon atom (group 4 in periodic table, i.e. 4 valence electrons) by an atom with 5 valenceelectrons, e.g. P, As, Sb, leaves one valence electron without a partner.Since the impurity contributes an excess electron to the lattice, it is called a donor.

SiSi

SiSi

SiSi

SiSi

SiSi

Si

Si

Si

P

Si

Si

Si

Si

ELECTRIC FIELD

PHOSPHORUS ATOM WITHNET POSITIVE CHARGE

NUCLEUS WITHCHARGE +5

EXCESS ELECTRON FROM PHOSPHORUS ATOM

Radiation Detectors and Signal Processing - II. Signal Formation Helmuth SpielerOct. 8 – Oct. 12, 2001; Univ. Heidelberg LBNL

The wavefunction of the dopant atom extends over many neighbors.


Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

Si

P

EXCESS ELECTRON FROM PHOSPHORUS ATOM


The excess electron is only loosely bound, as the Coulomb force is reduced by the dielectric constant εof the medium (ε =12 in Si).

2

( )( ) i

i

E atomE lattice

ε∝

The bound level of this unpaired electron is of order 0.01 eV below the conduction band (e.g. for P: Ec -0.045 eV).

⇒ substantial ionization probability at room temperature (E= 0.026 eV) – “donor”

⇒ electrons in conduction band

CONDUCTION BAND

VALENCE BAND

DONOR LEVEL

E


Conversely, introducing a group 3 atom (B, Al, Ga, In) leaves a Si valence electron without a partner.


To close its shell the B atom “borrows” an electron from a lattice atom in the vicinity.

This type of dopant is called an “acceptor”.

The “borrowed” electron is bound, but somewhat less than other valence electrons since the B nucleusonly has charge 3.

SiSi

SiSi

SiSi

SiSi

SiSi

Si

Si

Si

B

Si

Si

Si

Si BORON ATOM WITHNET NEGATIVE CHARGE

NUCLEUS WITHCHARGE +3

HOLE LEFT BY “BORROWED” ELECTRON

BORON ATOM “BORROWS”AN ELECTRON TO FILL ITSADJACENT VALENCE BONDS


This introduces a bound state close to the valence band, also of order 0.01 eV from the band edge.

For example, a B atom in Si forms a state at Ev + 0.045 eV.

Again, as this energy is comparable to kT at room temperature, electrons from the valence band can beexcited to fill a substantial fraction of these states.The electrons missing from the valence band form mobile charge states called “holes”, which behavesimilarly to an electron in the conduction band, i.e. they can move freely throughout the crystal.

Since the charge carriers in the donor region are electrons, i.e. negative, it is called “n-type”.Conversely, as the charge carriers in the acceptor region are holes, i.e. positive, it is called “p-type”.

CONDUCTION BAND

VALENCE BAND

ACCEPTOR LEVEL

E


pn-Junction

Consider a crystal suitably doped that a donorregion and an acceptor adjoin each other,a “pn-junction”.

Thermal diffusion will drive holes and electronsacross the junction.

Although the p and n regions were originallyelectrically neutral, as electrons diffuse from the nto the p region, they uncover their respective donoratoms, leaving a net positive charge in the n region.

This positive space charge exerts a restrainingforce on the electrons that diffused into the pregion, i.e. diffusion of electrons into the p regionbuilds up a potential. The diffusion depth is limitedwhen the space charge potential exceeds theavailable energy for thermal diffusion.

The corresponding process also limits the diffusionof holes into the n-region.

JUNCTION COORDINATE

x = 0

EFp

EFn

Vbi

FIXED CHARGE OF ATOMIC CORES

PO

TEN

TIAL


The diffusion of holes and electrons across the junction leads to a region free of mobile carriers – the“depletion region”, bounded by conductive regions, which are n- and p-doped, respectively.

Strictly speaking, the depletion region is not completely devoid of mobile carriers, as the diffusionprofile is a gradual transition.

Nevertheless, since the carrier concentration is substantially reduced, it is convenient to treat thedepletion zone as an abrupt transition between bulk and 0 carrier concentration.

Furthermore, the formation of the two adjacent space charge regions builds up a potential barrierbetween the n and p regions, which impedes the further flow of charge.

The magnitude of this potential barrier is typically 50 – 90%of the band-gap, depending on relative doping levels.

This represents the situation in thermal equilibrium. By application of an external potential, two distinctlydifferent non-equilibrium modes can be established.


a) positive potential applied to the p regionnegative potential applied to the n region

The externally applied voltage reduces the potential barrier, allowing increased charge transfer acrossthe junction.

⇒ “forward bias”

Electrons flowing from the n-region across the junction are replenished from the external voltage supplyand large current flow is possible.

p n

V

FORWARD BIAS

JUNCTION COORDINATE

x = 0

EFp EFn

POTE

NTI

AL


b) negative potential applied to the p regionpositive potential applied to the n region

This arrangement increases the potential barrier across the junction, impeding the flow of current.

⇒ “reverse bias”

Potential across junction is increased ⇒ wider depletion region

p n

V

REVERSE BIAS

JUNCTION COORDINATE

x = 0

EFp

EFn

POTE

NTI

AL


The p-n junction is asymmetric with respect to current flow (diode).

a) forward bias

positive supply connection → p contactnegative supply connection → n contact

⇒ large current flow

Diode current vs. voltage = −/0( 1)eq V kTI I e

(Shockley equation)

b) reverse bias

positive supply connection → n contactnegative supply connection → p contact

⇒ small current flow

-5 5

VOLTAGE (eV /kT)

CU

RR

EN

T (I

R/I

0)

0 1 2 3 4 5VOLTAGE (e|V|/kT)

0.1

1

10

100

CU

RR

EN

T (|

I R|/I

0)

10

5

-1


Since the depletion region is a volume with an electric field, it by itself could be used as a radiationdetector.

• The width of the depletion region is increased by reverse bias.

Depletion width and electric field in p-n junction

Assume a reverse bias voltage Vb and that the potential changes only in the direction perpendicular tothe n-p interface. Poisson's equation is then

d Vdx

Nqe2

2 0+ =ε

(1)

where N is the dopant concentration and qe the electron charge.

Consider an abrupt junction where charge densities on the n and p sides are Nd qe and Na qe,respectively.


If the limits of the depletion region are xn on the n-side and xp on the p-side, after two successiveintegrations one obtains on the n-side

dVdx

q Nx xe d

n= − −ε

( ) (2)

and

Vq N x q N xx

Ve d e d nj= − + +

ε ε

2

2(3)

where Vj is the potential at the metallurgical junction. For x = xn

V x Vq N x

Vn be d n

j( ) = = +2

2ε(4)

and the contribution of the n-region to the total reverse bias potential becomes

V Vq N x

b je d n− =

2

2ε. (5a)

Correspondingly, in the p-region

Vq N x

je a p=

2

2ε(5b)


and the total potential becomes

Vq

N x N xbe

d n a p= +2

2 2

ε( ) . (6)

Due to overall charge neutralityN x N xd n a p= (7)

and

Vq N

NN x

q NN

N xbe a

da p

e d

ad n= +

= +

21

212 2

ε ε. (8)

The depletion widths on the n- and p-side of the junction are

x Vq N N N

x Vq N N Nn

b

e d d ap

b

e a a d=

+=

+2

12

1ε ε

( / );

( / )(9)

and the total depletion width becomes

W x xV

qN N

N Nn pb

e

a d

a d= + =

+2ε. (10)


Detector diodes are usuallyasymmetrically doped. The startingmaterial (bulk) is lightly doped andthe junction is formed by diffusing orion-implanting a highly doped layer.

The external connection to the lightly doped bulk is made by an additional highly doped layer of thesame type (non-rectifying, “ohmic” contact).

• The depletion region then extends predominantly into the lightly doped bulk.

Other details:

The guard ring isolates the wafer edge (saw cut) from the active region.

In the gap between the detector electrode and the guard ring it is critical to provide a neutral interface atthe silicon surface to prevent formation of a conductive path.

This is best accomplished by oxide passivation (SiO2).

300 mµ~ 1 mµ

~ 1 mµ

GUARD RING

OHMIC CONTACT

JUNCTION CONTACTOXIDE

Si BULK


Strip and pixel detectors utilize a similar structure, except that the pn-junction issegmented:

p n-on- STRIPS n n-on- STRIPS

p-STRIP n-STRIP

n-SILICON n-SILICON

Al CONTACT STRIP INTERMEDIATE STRIPp-PSG

SiO SURFACEPASSIVATION

OHMIC CONTACT

2


When, for example, a dN N , the depletion region extends predominantly into the n-side and the totaldepletion width is

W xV

q Nnb

e d≈ =

2ε. (11)

The doping concentration is commonly expressed in terms of resistivity

ρ µ= −( )q Ne1,

because this is a readily measurable quantity. The parameter µ describes the relationship between theapplied field and carrier velocity (to be discussed later).

Using resistivity the depletion width becomes

W Vn n b= 2εµ ρ . (12)

Note that this introduces an artificial distinction between the n- and p-regions, because the mobilities µfor electrons and holes are different.

Since the mobility of holes is approximately 1/3 that of electrons, p-type material of a given dopingconcentration will have 3 times the resistivity of n-type material of the same concentration.


As discussed earlier, even in the absence of an external voltage electrons and holes to diffuse acrossthe junction, establishing a "built-in" reverse bias voltage Vbi. If we take this inherent bias voltage intoaccount and set for the bias voltage b b biV V V→ + , one obtains for the one-sided junction

.)(2)(21 bibnn

de

bib VVNq

VVxW +=+

=≈ ρεµε

For example, in n-type silicon (Vb in volts and ρ in Ω.cm): 0.5 x ( + )b biW m V Vµ ρ=

and in p-type material: 0.3 x ( + )b biW m V Vµ ρ=

The depleted junction volume is free of mobile charge and thus forms a capacitor, bounded by theconducting p- and n-type semiconductor on each side.

The capacitance is2( )

e

b bi

q NAC A

W V Vε

ε= =+

For bias voltages b biV V 1

b

CV

∝

In technical units1

1 [pF/cm]CA W W

ε= ≈

A diode with 100 µm thickness has about 1 pF/mm2.


The capacitance vs. voltage characteristic of a diode can be used to determine the doping concentrationof the detector material.

2( )e

b bi

q NCA V V

ε=

+

In a plot of (A/C)2 vs. the detector bias voltage Vb the slope of the voltage dependent portion yields thedoping concentration N.

Example: Si pad detector, A= 1 cm2, 100 µm thick2

12

1 (1/ ) 12 5 10

eqd CN dV

ε = = ⋅

0 10 20 30 40 50REVERSE BIAS VOLTAGE (V)

0

200

400

600

800

1000

CA

PAC

ITA

NC

E(p

F)

0 10 20 30 40 50REVERSE BIAS VOLTAGE (V)

0x100

2x1019

4x1019

6x1019

8x1019

1/C

2(F

-2)

Documents

Semiconductor Detectors and Electronics · Solid State Detectors and Electronics – Introduction Helmuth Spieler TRIUMF Summer Institute 2007 LBNL 3 When first introduced in the