Study of Silicon Photomultipliers Joëlle Barral, MPI, 25th June 2004

Study of Silicon Photomultipliers

Joëlle Barral, MPI, 25th June 2004

Joëlle Barral MPI 25th

June 2004

Study of Silicon Photomultipliers

A. Why SiPM : how to detect good detectors…?

B. From Avalanche PhotoDiodes to Silicon PhotoMultipliers

C. Some Features


June 2004

Why SiPM ?Or how to detect good

detectors…

Caran d’Ache Une planche qui regarde passer le train

High time resolution Short rise time Short recovery time

= FAST DETECTORS


June 2004

High precision Low noise rate Single photon resolution Efficiency

Additional features Low sensitivity to high magnetic / electric field

« low sensitivity to magnetic fields of the order of the gauss »…

Hadron calorimeter : 4T

Behaviour with temperature

Why SiPM ?Or how to detect good

detectors…


June 2004

From APD to SiPM…

Basic structure of an APDImpact ionization releasing EHPs and resulting avalanche multiplication

h+

E

šn+ p

e–

Avalanche region

S.O.Kasap, Optoelectronics

Geiger mode→binary device


June 2004

From APD to SiPM…

Pixels of the SiPM

SiPM

Silicon PhotoMultiplier (SiPM)MEPhI&PULSAR

Each pixel = binary device

SiPM=analogue detector

42 µm

20 µm1 mm

1 mm

24*24=576 pixels


June 2004

Electric field distribution in epitaxy layerTopology of SiPM

From APD to SiPM…Electrical decoupling

to readout the signal

Uniformity of the electric field


June 2004

Features

Energy Gain Single photon resolution Dynamic range

Noise Dark noise Afterpulse Crosstalk

Time Time resolution Rise time Recovery time

Parameters Overvoltage Temperature Light wavelength

(393 nm)→ enough?


June 2004

Gain vs overvoltage

Calibration on the dark noise (cross-talk)

Area on the scope (nVs)

19*1.6*10 *50C Geiger mode :

C = 36 pF

Gain~1.5 106 → low electronic noise

( APD Proportional mode : Gain~200 )

/ | | *( ) / | |pixel pixel pixel bias breakdowngain Q e C U U e


June 2004

Single Photoelectron Counting Poisson statistics?

54 V

A preamplifier is used

52 V 56 V

B. Dolgoshein Int. Conf. On New Developments in Photodetection, Beaune, France, 2002


June 2004

Limited Dynamic Range

• Saturation of the SiPM signal with increased light intensity(Average number of photoelectrons per pixel)

Number of photons arriving on the SiPM

Num

ber

of p

ixel

s fir

ed

Statistics=10 for each number of photons arriving

_

1*(1 (1 ) )photonsN

pixels firedN mm

m=total number of pixels=576

Joëlle Barral MPI 25th June 2004

0 200 400 600 800 1000 1200 1400 1600 1800 2000

700

600

500

400

300

200

100

0


June 2004

B. Dolgoshein The Silicon Photomultipliers in Particle Physics: Possibilities and Limitations

*

_ *(1 )photonsN

mpixels firedN m e

ε=photon detection efficiency




June 2004

_

1*(1 (1 ) )photonsN

pixels firedN mm

m=576

or*

_ *(1 )photonsN

mpixels firedN m e

?

1(1 ) 0.998264

m

1

576 0.998265e

…

Nu

mb

er

of

pix

els

fire

d


B. Dolgoshein An advanced study of Silicon Photomultiplier


1 1.4 1.8 2.2 2.6 3 3.4 3.8

Average number of photoelectrons per pixel

600

500

400

300

200

100

0


June 2004


Total number of pixels

Ave

rag

e n

um

be

r o

f p

ho

toe

lect

ron

s p

er

pix

el

20 photons firing

The increase of total pixel number seems technologically possible up to ~4000/mm ²

Hadron Calorimeter

- minimal signal 20 photons/mm²

- maximal signal 5000 photons/mm²

5000 photons firing

Ave

rag

e n

um

be

r o

f p

ho

toe

lect

ron

s p

er

pix

el

Total number of pixels

Irradiance of EUSO (clear sky conditions, primary proton E~1020eV, 45° zenith angle…) = 550 photons/m²


1

1.4

1

.8

2.2

2

.6

3

3

3

.8

0 400 800 1200 1600 2000 2400 2800 3200 3600 4000

1.7

2

.1

2.5

2

.9

3;3

3

.7

4.1

1200 1600 2000 2400 2800 3200 3600 4000


June 2004


Taking into account only the saturation of the pixels…

Nphotons>4056

576 pixels fired

10 %

1 %

Signal dispersionPessimistic…

B. Dolgoshein An advanced study of Silicon Photomultiplier

Relevant ?

22( )photons photonsN incert N

/ : 0.6 346photonsphotons

Nc c N

m

for Si, 400 nm ~70%

* *

*

geom Geiger

photoelectronsgeom

photons

QE

NQE

N

1 2 3 4 5 6Average number of photoelectrons per pixel


June 2004

Limited Dynamic RangeLess pessimistic…

Simulation

-Poisson statistics

-Saturation : incertitude in the number of photons detected

-Fluctuations around the saturation

photonsN

Statistics = 50

0 400 800 1200 1600 2000 2400 2800 3200

600

500

400

300

200

100

0

Number of pixels fired

12

10

8

6

4

2

00 400 800 1200 1600 2000 2400 2800 3200


June 2004


Statistics = 1000

2500 firing photons

10%

3700 firing photons

1/ photonsN


June 2004

Rise time

Ubias=56V

One-pixel amplitude~6 V

Rise time~1 ns

FWHM~2 ns


June 2004

Rise timeFHWM~2 ns

Ubias=56V

Rise time~1 ns

~500 pixels fired

APD :rise time=1ns


June 2004

Time resolution

Dependence with the number of pixels fired

Poisson statistics

One-pixel time resolution

σ=17ps

FWHM = 402 ps

σ=171ps

Best time resolution

( 27 ps )

Oscilloscope time resolution

Picosecond Pulsed Diode Laser PDL 800-B :

• Synchronisation Output < 20 ps

Electronics noise

σ/√2=7 ps

Traps in deep levels

_

1

pixels firedN


June 2004

Time resolution

Randomness in physical mechanisms : ultimate limits

Photon absorption in the depletion layer Distance point of absorption / High field region Depth of the depletion layer Position over the active area : transverse propagation of

the avalanche activation (lateral drift and diffusion of free carriers)

Avalanche multiplication = stochastic processFluctuation (number, position) of ionizing events


June 2004

Recovery time Quenching

Passive

Active


S. Cova et al. Evolution and Prospect of Single-Photon Avalanche Diodes and Quenching Circuits

Difficult for each pixel


June 2004

Recovery time Diode model

Dependence of the overvoltage

1t

RCe

1.2 µs but…Joëlle Barral MPI 25th

June 2004

RpixelCpixel=400 kΩ *36 fF ~ 15 ns

all pixels fired


June 2004

Recovery time

12

12001 1%e

40 ns ?

( It’s bad…)Ubias=56V


June 2004

Recovery time All pixels fired

Limits of the dynamical range Recovery time τ of one pixel

→ τone pixel=1.2 µs

Some pixels fired ?

*(1 )t

m e

Recovery time for one pixel


June 2004

Recovery time

Signal detected (normalized / number of pixels fired)

Delay t between the two firing signals (ns)

63% of maximal value

2*[1 (1 1/ ) ]*(1 1/ ) *[1 (1 1/ ) ] *(1 )photons photons photons

tN N Nm m m m m e

Recovery time = 119 ns

formulasimulation

0 1000 2000 3000 4000 5000

250

230

210

190

170

150

130

Example with two firing signals of 300 photons


June 2004

Recovery time

Number of photons firing ( >264)

Recovery time (ns)

1 *(1 ln[1 (1 1/ ) ])photons

photons

NN pixel m if

>0…

N photons firing = 815

N pixels fired = 436

63%

200 600 1000 1400 1800 2200 2600 3000

1200

1000

800

600

400

200

0


June 2004

Dark noise

Electron-hole recombinations / Carrier generations

Optical electron-hole pair generation

Thermal electron-hole pair generation

Impact ionization

Theoretically impossible in indirect semiconductor

Dark counting rate

@ room temperature

~1 MHz


June 2004

Afterpulsing Time Correlated Carrier Counting

θ=dark-noise rate Trapping levels

* ( )dN

K N tdt

*( ) * j

t

tj

j

P t B e A e


June 2004

Afterpulsing

Hold-off time = 3.4µs

τ1=141 ns τ2=289 ns

τ3=155 ns τ4=393 µs

Probability<10%

Dark counting rate56 V : 1 MHz →1/θ=1 µs


June 2004

Crosstalk

Trenches

1 pixel : 76%

2 pixels : 18% 3 pixels :

5% 4 pixels 1%

1

2

3

Hot carrier luminescence : 105 avalanche carriers→1 photon emitted

1. Direct cross-talk2. Inside the depletion layer3. Through reflection

Dolgoshein Status of upgrade SiPM developments

1 pixel

3 pixels

2 pixels


June 2004

Application : Positron-Emission Tomography 22Na decay : β+ emission Annihilation radiation Coincidence measurement

ep e n

2e e

22Na

γ

511keV

γ

511keV

LSO scintillator 2mm*2mm*10mm

SiPM 1mm*1mm

enough?


June 2004

Application : Positron-Emission Tomography

PET for brain MPI für neurologische Forschung, Köln

Philips, PET, Allegro


June 2004


Compton scattering interaction

Photopeak

Energy windows around the 511 keV photopeaks to :• reduce the chance of fortuit coincidence • cut the spatial dispersion (Compton)


June 2004


σ=1.3 ns

S.R. Cherry Planar APD Arrays for High Resolution PET

σ=2.04 ns

4*4 APD coupled to 2*2*10 mm LSO arrays

Pichler Entwicklung eines Detektors für die hochauflösende PET(…)

σ=1.4 ns

2*8 LSO-APD matrix


June 2004

QUESTIONS ?

Documents

Study of Silicon Photomultipliers Joëlle Barral, MPI, 25th June 2004