Introduction to Silicon Detectors

G.VillaniSTFC Rutherford Appleton Laboratory

Particle Physics Department

Outlook

• Introduction to physics of Si and detection– Si electronic properties, transport mechanisms, detection

• Examples of detectors– Strips, CMOS,CCD,MOS

• Radiation damage• Conclusions

2

Introduction

The Si detection chain

3

Sensing/Charge creation

Charge transportand collection

ConversionSignal

processingData TXE

Si physical properties

Si device properties Si device topologies properties

almost all the boxes of the detection chain process based upon Silicon

The crystalline structure is diamond cubic (FCC), 8 atoms/cell with lattice spacing of 5.43 A ~ 5x1022 cm-3

* In electronic industry all crystallographic forms are used (Single crystal, Polysilicon, α-Si)

The key to success of Si is related to its abundance and oxide SiO2, an excellent insulator (BV ~ 107 V/cm).

* Micro crystals but the flexible bond angles make SiO2 effectively an amorphous: its conductivity varies considerably (charge transport in SiO2 via polaron hopping between non-bonding oxygen 2p orbitals)

Silicon properties

After Oxygen, Silicon is the 2nd most abundant elementin Earth’s crust (>25% in mass)

Si

1.48A

4

Silicon electrical properties

5

* The appearance of Band Gap, separating CB and VB

* The 6 CB minima are not located at the center of 1st Brillouin zone, INDIRECT GAP

CB VB-H VB-L

1st Brillouin zone of Diamond lattice

CB

VB

Silicon Band structureThe electronic band structure can be obtained within the independent electron approximation (normally 1 electron SE in periodic potential neglecting electron interactions) in terms of Bloch functions

kEerurEUT nrjk

knkn ,,~ a wave associated with free motion of electrons modulated by the periodic solution u n,k. The energy E is periodic in k so is specified just within the 1st unit WS cell of the reciprocal lattice (the Brillouin zone).

Silicon electrical properties

The detailed band structure is complicated: usually quasi-equilibrium simplifications are sufficient to study the charge transport.Assuming that the carriers reside near an extremum, the dispersion relationship E(k) is almost parabolic:

*0

*0

*0

22 1

2 m

p

m

kkEv

m

kkE k

FV

dt

pd

dt

tdkr

* Under the assumptions of small variation of the electric field, the carrier dynamics resembles that one of a free particle, with appropriate simplifications.* The effective mass approximation takes into account the periodic potential of the crystal by introducing an effective carrier mass ( averaged over different longitudinal and transverse masses). The lower the mass, the higher mobility (µ 1/m*)

* Similar approach used to calculate the E(k) for phonons.

oEE

mEg

m

kkE

32

2/3*0

*0

22

2

2

2

6

(3D)

Silicon electrical propertiesThe carrier density is calculated from:

• The density of states g(E), which depends on dimension;• The distribution function F(E);

Only partly filled bands can contribute to conduction: carrier density in CB and VB.At equilibrium the carrier density is obtained by integrating the product:

The density of states gD(E) depends on the dimension

ii

kTEvEcVCDD pneNdEEFEgn /

/

kTEE

EFFexp1

1

3

2

1

0In intrinsic Si a creation of e in CB leaves behind a hole in VB, that can be treated as an e with positive charge and mobility of the band where it resides

CB

VB

Fermi level: energy level @ 50% occupancy

7

KTeNNnpn

kTE

VCi

g300@10

20/2

Conduction of Si intrinsic @ T = 300K:

σ = q(μn +μp) ni = 3.04x10-6mho-cm ->329kOhm-cm

By adding atoms of dopants, which require little energy to ionize( ~10’s mEV, so thermal energies @ ambient temp is enough) we can change by many odg the carrier concentration.

Doping concentration: 1012 to 1018 cm-3

In crystalline Si ~ 5*1022atomscm-3

In equilibrium and for non degenerate casethe relationship between carrier concentration and E is the same as in the intrinsic case:

32

17

20/2

1010:..

300@10

D

iDD

kTE

VCi

N

nppNpnNge

KTeNNnpn g

8

Silicon electrical properties

Charge transport

Charge transport:The charge transport description in semiconductors relies on semi-classical BTE (continuity equation in 6D phase space)

k

k

k

coll

krk

tkrfkEV

trW

tkrfkvV

qtrJ

tkrfV

trn

tkrSt

tkrftkrf

FtkrfkE

t

tkrf

,,1

,

,,,

,,1

,

,,,,

,,,,1,,

The distribution function f(r,k) can be approximated near equilibrium:

equilibrium Near equilibrium

k0

fcoll

ff

t

tkrf

0,,

Q conservation

P conservation

E conservation

9

Charge transport

Under (many) simplifying assumptions the 1st moment of BTE gives the DD model(The semiconductor equations):

AD

pp

nn

ppp

nnn

NNnpV

UJqt

p

UJqt

n

pqDrEqpJ

nqDrEqnJ

1

1

Transport of charge is a combination of drift and diffusion mechanism

* DD expresses momentum conservation: it becomes invalid when sharp variation in energyof carriers occur (due to F for example: deep submicron devices)When feature size is 0.’sμm the DD model becomes invalid: higher momentum required

Diffusion termDrift term

10

Detection principles

A: Ionization: by imparting energy to break a bond, electrons are lifted from VB to CB then made available to conduction ( ionization chambers, microstrip, hybrid pixels, CCD, MAPS…)

Bethe formula for stopping power gives the rate of energy loss/unit length for charged particles through matter

Photon interaction zEoeIzI

α

MIP

11

Detection

nmI

vR

cmRdx

dEn

i

110

10311 315

2

zEoeIzI

A MIP forms an ionization trail of radius R when traversing Si, creating ~ 80e-/μmLow injection regime:the generated charge is too small to affect the internal electric field

MIP charge density

cmL

cmEm

h

7

15

1015.0

102

The associated wavelength is much smaller than mean free path:Each charge is independent from each other;Carrier dynamics does not need QM

meze

h

Pzn in

/106.5 6

Photoelectric charge density

An optical power of -60dBm (= 1nW) of 1keV photons generates ~ 6*106e-/μm

High injection regime:Plasma effects The internal electric field can be affected by the generated charge

12

Detection

B: Excitation: Charge or lattice (acoustic or optical phonons) some IR detectors, bolometer

Dispersion relation for phonons in SiPhonon excitation energy ~ 10 meV : much lower threshold

60meV

Ec

EF

EVEF

~10’s meV

Eigenvalues separation in quantized structures ~ 10’s meV

SiSiO2Poly Si

13

Signal conversion: The pn junction

Homojunction: consider two pieces of same semiconductor materials with different doping levels:In equilibrium, the Fermi level equalizes throughout the structureThe thermal diffusion of charge across the junction leaves justthe ionized dopants : an electric potential, and a field F, develops across the junction

t

t

Vpp

Vnn

epppqDrEqp

ennnqDrEqn

0

0

0

0

In equilibrium J = 0: using DD model

ASCE (Abrupt Space Charge Edge) approximation:A ‘positive’ voltage increases (exponentially) the charge concentration: high direct current.A ‘negative’ voltage decreases it (down to leakage): the current reduces and at the same time widens the depleted region.Unidirectionality of current characteristics

Near the interface, the carrier concentration exponentially drops: a depletion region (empty of free charge) is formed.

0

14

Signal conversion: The pn junction

The electric field F in the depletion region of the junction is sustained by the ionized dopants. When charge is generated is swept across by the field

PN junction signal converter: A capacitor with a strong F across

A device with a large depleted region W can be used to efficiently collect radiation generated charge ( Solid state ionization chamber)

da

dab

NN

NN

q

VW

2W

To achieve large W high field region:• Low doping (high resistivity) Silicon is

needed• Large voltages

Conversion: Q to V // Q to I

15

Detectors examples

Strip detectors

Scientific applications

Monolithic Active Pixel Sensors (MAPS)

Imaging, consumer applications

Charge Coupled Devices (CCD)

Imaging, scientific and consumer applications

MOS detector

scientific applications

RAL PPD has (is) actively involved with all these detector technologies

16

Detectors examples

Use of Si Strip detectors

Almost all HEP experiments use Si detectors:The high density track region usually covered by pixel detectors; by strip at larger radius (cost reason)

17

Detectors examples

ATLAS SCT

4 barrel layers,2 x 9 forward disks

4088 double sided modules

Total Silicon surface 61.1m²

Total 6.3 M channels

Power consumption ~ 50kW

Events rate: 40MHz

Put stave pics of AUG!

768 Strip Sensors

RO

module

18

Detectors examples

Array of long silicon diodes on a high resistivity silicon substrateA strong F in the high resistivity Si region helps collect charge efficiently (drift).The transversal diffusion of charge implies a spread of signal over neighbouring strips

The high resistivity Si is not usually used in mainstream semiconductor industry:

Hybrid solution: detectors connected (wire/bumpbonded) to the readout electronic (RO)

P++

N+ (high res)

Vbias ~100’sV

F

Strip detectors 768 Strip Sensors

RO electronic

Power supply

80μm

30

0μ

m

Wires

19

Detectors examples

High events rate require fast signal collection:Estimate of charge collection time in strip detector:

z

zo

z

z

dzF

Wz

Fdz

zvzt

0011

111

For a detector thickness of 300um and overdepleted Vb = 50V and 10kohm resistivity

tcoll(e)≈ 12ns tcoll(h)≈ 35ns

The fast collection time helps the radiation hardness:The radiation damage to sensors is a crucial issue in modern HEP experiments

20

Detectors examples

RO electronic

3T ( 3MOS) MAPS structure

2D array of ~106 pixelsMonolithic solution:Detector and readout integrated onto the same substrate

≈10’s m

RO

electronic

MAPS detectors

21

Detectors examples

MAPS detectors

The charge generated in the thin active region moves by diffusion mainly:‘Long’ collection timeSmall signal

Different implants arrangements for charge collection optimizationCircuit topologies for low noise

N++ (low res)

P++ (low res)

P+ (low-med res)

Vbias ~V’s

Mechanical substrate100’s μm

Active region‘s μm

Electronics0.’s μm

22

Detectors examples

Charge collection time (s) in MAPS vs. perpendicular MIP hit

10-7

TPAC 1 pixel size 50x50um2Chip size ~1cm2

Total pixels 28k>8Meg Transistors

n

collnn D

ltUnD

t

n 22

Example of MAPS detectors:

23

Detectors examples

Once the charge has been generated, it accumulates in the potential well, under the capacitor.The control circuitry shifts the accumulated charge to the end of the row, to the input of a charge amplifier. The sensor is fabricated in a optimized, dedicated process and the RO on a separate chip. Superior imaging quality but less integration and speed.

Nobel Prize 2009 for Physics toinventors Boyle and Smith

CCD detectors

24

Detectors examples

In-situ Storage Image Sensor: ISIS

CCD in CMOS process 0.18μmCharge collection under a PG then stored under a 20 pixels storage CCD

RG RD OD RSEL

Column transistor

On-

chip

logi

c

On

-ch

ip s

witc

he

s

Global Photogate and Transfer gate

ROW 1: CCD clocks

ROW 2: CCD clocks

ROW 3: CCD clocks

ROW 1: RSEL

Global RG, RD, OD

Imaging pixel

5 mm

80 mm

55Fe g source Mn(Ka)

Mn(Kb)

25

Signal conversion: The unipolar MOS device

By applying a voltage to the G with respect to the Substratean electric field develops across the SiO2: a charge channel is formed between Source and Drain. The Ids characteristics depends on the Vgs applied.

The CMOS process refers to the minimum feature sizeachievable i.e. the channel length)Currently 45nm: the modelling of the characteristics of the device of this size is non-trivial: • Quantization effects at the boundary;• QM tunnelling across the gate;• Hot carriers near the D/S junction;• …

Metal Oxide Semiconductor device are unipolar devices based on voltage modulation of charge.The control gate is physically separated by the active region where the charge moves by a thin (nm) layer of SiO2.

P++

N++

SiO2NMOS

26

LET in SiO2 for different particles

0 0 1

)()(1

!!),,(

m n nml

lmKA

oox l

K

m

AeeKrTFY

kT

qr

r

rA

SiO

c

o

c

24

,2

kT

rqFK oox

Generation rate in SiO2 vs. electric field

The SiO2 is a very good insulator: a strong electric field can be applied to it and the charge generated in SiO2 by ionizing radiation efficiently collected

However SiO2 is a polar material: the recombinationprocesses are stronger than in Si. Furthermore, holeTransport is non Gaussian (low ‘mobility’) and traps form near Si interface.

27

Signal conversion: The unipolar MOS device

Addition of a Floating Gate (FG): the electrical characteristics of the device are controlled by the chargestored in the FG. The electric field in the SiO2 due to the FG drifts charge towards/away from it.The discharge of the FG alters the device electrical characteristics

FGSiO2

FGSiO2

Floating Gate

Control Gate

Conversion: Q to I

Pre-rad

Post-rad

Reprog

Ids(A)

Chip #1 100Gy

<∆Vth > 0.6152

Std dev 0.00598

Radiation sensitivity

The MOS structure easily allows

excitation based radiation detection

28

SiO2

Signal conversion: The unipolar MOS device

Radiation damage

NIEL/cm2/yrTID/Gy/yr ATLAS

Total Ionizing Dose (rad = 0.01Gy) Non Ionizing energy Loss (1MeV neutrons/cm2 fluence)

In HEP and space applications the detectors are exposed to high level of radiation:LHC: 10’s Mrad (100kGy) over 10years of operation

N.B.: 1 rad/cm3 Si ~1013e/h pairs

29

Radiation environment in LHC experiment

TID Fluence 1MeV n eq. [cm-2] @ 10 years

ATLAS Pixels 50 Mrad 1.5 x 1015

ATLAS Strips 7.9 Mrad 2 x 1014

CMS Pixels ~24Mrad ~6 x 1014 CMS Strips 7.5 Mrad 1.6 x 1014

ALICE Pixel 250 krad 3 x 1012

LHCb VELO - 1.3 x 1014/year

All values including safety factors.

30

Radiation damage

Microscopic effects: Bulk damage to Silicon : Displacement of lattice atoms (~ Kinetic Energy Released)

Atoms scattered by incoming particles leave behind vacancies or atoms in interstitial positions (Frenkel pairs).Low energy particle ~ point defectsHigh energy particles ~ cluster defects

V

I

Vacancy + InterstitialEK>25 eV

31

Radiation damage

Energydeposition

Atoms displacement

alteredLattice

periodicity

Band gap Spurious

states

The appearance of spurious band gap states affects the electro/optical characteristics of the device:

• Thermal generation of carriers (increased leakage current @ same T)• Reduced recombination time ( quicker charge loss , reduced signal)• Charge trapping• Scattering• Type conversion

AlteredElectrical

characteristics

Conduction band

Valence band

Band gap

+++Donor levels

Acceptor levels

generation recombination

-compensation

trapping

32

Radiation damage

Detrimental Macroscopic effects:• Noise increases because of increase leakage current• Charge Collection Efficiency (CCE) is reduced by trapping• Depletion voltage increases because of type inversion

1015 1MeV n-eq.

tQtQ

heeffhehe

,,0,

1exp)(

defectsheeff

N,

1

33

Radiation damage

To increase the Radiation Hardness of Sensors:• Operating conditions (cooler – lower leakage)• Material engineering ( OFZ - Diamond detectors)• Device engineering (n in n – 3D detectors)

Electrodes in the bulk – lateral collectionThe device achieve full depletion• Low depletion voltage• short collection time• claim reduction in signal 33% after 8.8X1015 1Mevn• difficult to manufacture

• 3D DDTC similar to 3D but easier to manufacture; alsobetter mechanical strength.

* Radiation damage affects also the RO electronics, but modern process can address the problem efficiently( guard rings, sub micron devices)

34

Radiation damage

Addendum - Detector systems

The ATLAS SCT (semiconductor tracker) detector. The thick red cables on show feed the detector with half of its power – adding more will take up even more space

HEP experiments: large detector systemsChallenging engineering issues

35

ALICE ATLAS CMS LHCb

Strips 4.9m2 64m2 210m2 14.3m2

Drift 1.3m2

Pixels 0.2m2 2m2 1m2 0.02m2

Number of Channels

Strips 2.6 x 106

6.3 x 106

9.6 x 106

1 x 106

Drift 1.3 x 105

Pixels 9.8 x 106

80 x 106

33 x 106

1 x 106

Addendum - Detector systems

A serial powering or DC2DC approach can increase efficiency in power distribution compared to a parallel approach

SP

DC2DC

Alternative powering schemes:

ATLAS SCT Barrel 3 at CERN. Half of the 384 cables are visible; the rest enters the other end of the detector.

36

The field of semiconductor detectors encompasses different scientific and technology fields: solid state physics, nuclear and particle physics, electrical engineering, …

Some of the issues relevant to radiation detectors:

• Radiation hardness • Topologies optimization (power reduction, noise reduction)• Development of new detection techniques based on novel and well established

semiconductor material: ( phonon-based detectors, compounds, low dimensional)• Integration with electronics (monolithic solution to achieve more compactness

and reduce cost),3D structures

Conclusions

37

Backup - Detector systems

=

At pixel level, power consumption could be optimized by using a non linear approach:

The positive feedback structure is biased near threshold (variable)

A small signal triggers the structure

Power reduction at detector level

I

Backup - Detection

Intrinsic resolution of Si and Ge based detectors

The variance in signal charge σi associated to the ionization process is related to the phonon excitation

1

i

i

i

pn

i

oi EE

EE

High resolution requires smaller band gap (εi ), direct or small phonon excitation energy

Fano factor ~0.1 in Si

II

Backup -Detection

The indirect BG of Si requires higher energy for charge excitation, because energy and momentum must be conserved (Phonon-assisted pair creation/recombination)

In Si an average of 3.6 eV is required for pair creationPut values of photon momentum typ.

Ph: DQ~107 m-

1

DQ~1010 m-1

/p a

III

Backup

Quantization effects due to band bending in Si-SiO2 interface: excitation based detection

Si-polySi-sub

SiO2

Q-effects

IV

Backup - The bipolar transistor device

A bipolar transistor can be thought of as a two diode system, connected in anti series;• One is forward biased;• The other is reverse biased

The bipolar transistor can be (and it is) used as a high gain detector

Main limitations arising from speed: the minority carriers diffuse through the base ( relatively low speed)

V