Physics and Experimental Studies of SiPM Nonlinearity and Saturation - - PowerPoint PPT Presentation

June 14 2018 E.Popova SiPM SiPM Nonlinearity... 1

Physics and Experimental Studies of SiPM Nonlinearity and Saturation

• Dr. Elena Popova

National Research Nuclear University MEPhI, Moscow, Russia

14th June 2018 Schwetzingen, Germany ICASIPM– the International Conference on the Advancement of Silicon Photomultipliers

E.Popova SiPM SiPM Nonlinearity... 2 June 14 2018

1 10 100 1000 10000 1 10 100 1000 Number of pixels fired Number of photoelectrons 576 1024 4096

⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ − ⋅ =

⋅ −

total photon

N PDE N total firedcells

e N N 1

dark low intensity light Light (single photon) Silicon Photomultiplier – has been developed for single photon applications

• V. Andreev et al. / NIM A 540 (2005) 368–380

For short light pulses due to finite number of cells charge signal (counted in number of fired cells) saturates Sometimes people use it for very high intensity light registration Example : Calorimetry BUT Binominal approach:

E.Popova SiPM SiPM Nonlinearity... 3 June 14 2018

A Monte-Carlo model of a SiPM coupled to a scintillating crystal 2012 JINST 7 P02009 (http://iopscience.iop.org/1748-0221/7/02/P02009) Saturation SiPM signal

E.Popova SiPM SiPM Nonlinearity... 4 June 14 2018

E.Popova SiPM SiPM Nonlinearity... 5 June 14 2018

E.Popova SiPM SiPM Nonlinearity... 6 June 14 2018

Shaojun Lu

If we are studying SiPM properties we have to think in the coordinates of

• fired pixels (together with correlated pixels) – Y
• Number of phe assuming ideal conditions with infinite number of pixels inside SiPM) – x

E.Popova SiPM SiPM Nonlinearity... 7 June 14 2018

How to normalize the SiPM saturation curve?

We need several Light intensity points provide us enough value of P(0)

• 0,5

0,0 0,5 1,0 1,5 2,0 2,5 3,0 50 100 150 200 250 300 Events Area, nVs

Maybe last intensity calibration point (too low P(0))

<Nphe>≈5

Amplifier and SPE spectra (low intensity light)

K1e

Mean= Mean(whole distribution)-Ped_position

< N phe >= −lnP(0)

e pixels fired

К Mean N

1 _

>= <

Xi= Yi = Zi = signal from the reference photodetector, units

0,1 1 10 100 1000 10000 0,1 1 10 100 1000

wafer #5 5 SiPMs

K type (1024 pixels)

Number of photoelectrons Number of fired pixels

phe Fired pixels

5 SiPM samples SiPM Crosstalk is visible Fired pixels>phe

20 40 60 80 100 5 10 15 20

SiPM seed reference photodetector Z, units

Calibration of reference photodetctor in number of phe’s

calibration Only reference photodetector

Extraction number of seeds

E.Popova SiPM SiPM Nonlinearity... 8 June 14 2018

50Ω How to study the SiPM saturation?

• 1. Firstly we need to have a proper experimental setup
• 1. Light source, operated in stable mode (no changes in an electrical pulse)
• 2. Light intensity is changed by filters
• 3. Uniformly distributed light over the SiPM surface* (over surface with desired number
• f investigated pixels)
• 4. Reference stable linear photodetector (the best choice is PIN-diode)
• 5. Amplifier to obtain SPE spectra for low light intensity (bypassed for high intensity

light)

• 6. Temperature and voltage must be stable and better controlled with needed accuracy

Example of the setup

E.Popova SiPM SiPM Nonlinearity... 9 June 14 2018

The proper experimental setup

• 1. Light source, operated in stable mode (no

changing of electrical pulse)

• 2. Light intensity is changed by filters

Due to changing of electrical pulse light pulse shape, wavelength and distribution of correlated photons might be changed too 3.Uniformly distributed light over the SiPM surface* (over surface with desired number of investigated pixels) Saturation (nonlinearity) depends on pixel load (number of photons/number pixels (think in fraction)

• 4. Reference stable linear photodetector -the best choice is PIN-diode

(dynamic range of about 108) PMT is not the best choice for the reference detector. It has own nonlinearity, especially for pulse signals (parameters of the specific PMT should be checked)...

E.Popova SiPM SiPM Nonlinearity... 10 June 14 2018

7.Then we need to understand what we want to study– amplitude (A) or charge (Q)? A and Q are different and have a different behavior

• 8. Important – we need to know exactly pulse shape

corresponded to our task and the best situation when it can be reproduced exactly in the test setup

• 9. We need to know real operation conditions of SiPM

(applied voltage, light distribution over the SiPM area, load and serial resistances of a connection scheem)

E.Popova SiPM SiPM Nonlinearity... 11 June 14 2018

Amplitude (A) or Charge (Q)?

Before saturation doesn’t matter. But if you have more then one phe/pixel:

E.Popova SiPM SiPM Nonlinearity... 12 June 14 2018

Focused laser light at the center of the cell, 40ps, 660nm Scope LeCroy WaveRunner 620Zi 2GHz

25,5 26,0 26,5 27,0 27,5 28,0 28,5 29,0

• 8
• 6
• 4
• 2

Amplitude, mV time, ns

Why so significant dispersion of signals amplitudes? It is exactly one fired cell (stand alone) Suggestion – Geiger discharge starts from several points inside of the cell Single stand alone cell. Moderate light (several photons/flash) intensities MEPHI cell

E.Popova SiPM SiPM Nonlinearity... 13 June 14 2018

1 2 3 4 5 6 7 8 100 200 300 U =35V (U bd=33,35) N pixel=0,069 N pixel=1,74 N pixel=3,62

Events Area, pVs

Charge (exp)

2 4 6 8 10 12 14 200

Events Amplitude, mV

U =35V (U bd=33,35) N pe=0,069 N pe=1,74 N pe=3,62

Amplitude (exp) 1 2 3

E.Popova SiPM SiPM Nonlinearity... 14 June 14 2018

But what is about cell charge for high intensity light in reality?

E.Popova SiPM SiPM Nonlinearity... 15 June 14 2018

Pulse shape for different light Intensities. MEPHI data Hamamatsu S10362-11-100U No.50, Ubreakdown=68.4V, U=69.5V

10 15 20 25 30 0,00 0,05 0,10 0,15 0,20 Amplitude, V time, ns

from 0.4 phe/cell to 30k phe/cell

10 15 20

0,00 0,02 0,04 0,06 0,08 0,10 0,12 0,14

relative amplitude, a.u. time, ns Normalized pulse shapes Fast part Slow part (recovering)

E.Popova SiPM SiPM Nonlinearity... 16 June 14 2018

3.Important – we need to know exactly pulse shape corresponded to our task and the best way – it should be reproduced exactly in the experimental setup

1 10 100 1000 10000 1 10 100 1000

100 1 10

Energy Deposited, MIP

25 1

SiPM signal Number of phe

Individual tile energy reconstruction using calibration curve SiPM signal vs energy deposited:

200 400 600 800 1000 1200 200 400 600 800 1000 1200 1400 1600 1ch = 50 ps

LED Tile

TDC channel Counts

50 100 150 200 250

Counts

,pixels

0,1 1 10 100 1000 10000 0,1 1 10 100 1000

wafer #5 5 SiPMs

K type (1024 pixels)

Number of photoelectrons Number of fired pixels

phe Fired pixels

5 SiPM samples SiPM Crosstalk is visible Laser pulse FWHM 40ps

The same SiPM type 1024 pixels in saturation ~2000 pixels in saturation Pulse shape depended – recovery during pulse duration!!!

E.Popova SiPM SiPM Nonlinearity... 17 June 14 2018

E.Popova SiPM SiPM Nonlinearity... 18 June 14 2018

CALICE MINICAL (preprototype of the tile HCAL) 100 SiPMs individually read out tile+WLS 1024 real pixels inside (agrees with saturation curve for 40ps light) ⎥ ⎥ ⎦ ⎤ ⎢ ⎢ ⎣ ⎡ − ⋅ =

⋅ −

total photon

N PDE N total firedcells

e N N 1 Effective Ntotal is 1650+-150

E.Popova SiPM SiPM Nonlinearity... 19 June 14 2018

SiPM Recovery

Double light pulses method. 2 short pulses with high intensity to fire all SiPM cells Uniforme illumination over SiPM area y(Δt)=A2/A1 Fixed intensity Δt A1(Q1) A2(Q2) But one should be carefull – recovery might depends on light intensity (pixel load) - oversaturation

E.Popova SiPM SiPM Nonlinearity... 20 June 14 2018

0,0 0,2 0,4 0,6 0,8 1,0 1,2 1,4 1,6 1,8 2,0 2,2 20 40 60 80 100 120 140

smaller LIGHT Chi^2/DoF = 0.12033 R^2 = 0.99987 y0 138.12579 ±0.06337 A

• 138.72352

±0.05169 tr 0.87725 ±0.00121 LIGHT MAX Chi^2/DoF = 0.30513 R^2 = 0.99971 y0 153.60823 ±0.18718 A

• 154.82526

±0.13935 tr 1.23323 ±0.0034

C harge integration 25 ns

C harge of s econd puls e, Q D C channels

time between pulses, µs

y = A*exp(-x/tr) + y0

SiPM recovery time. Pulse shape analysis and double light pulses method for charge Q

Both methods give the same results for recovery time vs light intensity Drawback – no light intensity monitor Light MAX τr=1233±4ns Smaller light (but still with SiPM saturation) τr=877±2ns

E.Popova SiPM SiPM Nonlinearity... 21 June 14 2018

We have repeated our measurements with 1x1 mm2 MEPHI SiPM (pitch 100µm) under control of light intensity

1000 2000 3000 4000

250 500 750 1000

Time constant τ, ns

Nphel / cell

Amplitude, Charge and τ increase with light intensity Question – what is (are) the reason(s) for this? SiPM gain ~107 Group effect? U is fixed Slow part (recovery) SiPM pulse for saturation conditions SiPM recovery time τ Fast part (geiger discharge)

E.Popova SiPM SiPM Nonlinearity... 22 June 14 2018

Single cell pulses for high intensities light (fixed voltage U=35V). MEPHI cell (100x100µm2)

0,0 0,5 1,0 1,5 2,0 5 10 15 20 25 30 35 40

U bd = 33,25±0,05

Amplitude, mV time, ns

τ=1009ns ( Nphel=4,4) τ=1755ns (Nphel=9200) τ=2493ns (Nphel=15200) τ=2435ns ( Nphel=18800)

800 1600 1E-5 1E-4 1E-3 0,01

Amplitude, V time, ns

τ=1009ns ( Nphel=4,4) τ=2435ns ( Nphel=18800)

) ( * ) (

br slow fast

U U C C Q − + =

) ( *

slow fast quench

C C R + = τ

Total charge Fast part Slow part

Cfast – readout (parasitic) cell capacitance Cslow – cell p-n junction capacitance

SiPM cell pulse for saturation conditions

E.Popova SiPM SiPM Nonlinearity... 23 June 14 2018

Single cell pulses for high intensities light (for fixed voltage). MEPHI cell (100x100µm2)

const R U U Q

quench br

= − = / ) ( /τ

5 10 15 20 2 4 6 8 10 Q/τ, a.u

x103 Nphel

U-Ubreakdown doesn’t change Q increases due to increasing of Cfast+Cslow

const U U

br =

−

Specific technology?

5 10 15 20

1 2 3

U=35V , T =+23

0C

U bd = 33,25±0,05 (T =+23,5±1

0C )

Relative Charge Q,a.u.

x10

3 Nphel

) ( * ) (

br slow fast

U U C C Q − + =

Charge

5 10 15 20

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 T =+23

0C

U bd = 33,25±0,05 (T =+23,5±1

0C ) U=35V

Time constant τ, µs x10

3 Nphel

) ( *

slow fast quench

C C R + = τ

Recovery time

E.Popova SiPM SiPM Nonlinearity... 24 June 14 2018 24

0,0 0,6 1,2 0,0 0,1 0,2 0,3

0.397 phe 3.452 phe 5420 phe 19680 phe

F B K . UV S P AD 30 um. T =+20

0C

U bd = 26,06±0,04

Amplitude, V time, ns

400 800 1E-4 1E-3

0.397 phe 19680 phe

F B K . UV S P AD 30 um. T =+20

0C

U bd = 26,06±0,04

Amplitude, V time, ns

Very small difference in pulse shapes for different light intensities ΔU=12V Single cell pulses for high intensities light (for fixed voltage U=38V). FBK UV

• SPAD. Dia 30 µm

Thanks to F.Acerbi SiPM cell pulse for saturation conditions

E.Popova SiPM SiPM Nonlinearity... 25 June 14 2018

Single cell pulses for high intensity light (for fixed voltage U=38V). FBK UV SPAD. Dia 30 µm

10 20 30 40 20 40 60 80

Q/τ, a.u x10

3 Nphel

U-Ubreakdown doesn’t change Q increases due to increasing of Cfast+Cslow ΔU=12V

const R U U Q

quench br

= − = / ) ( /τ

10 20 30 40 50 60 0,4 0,6 0,8 1,0 1,2 1,4

relative charge Q, a.u. x10

3 Nphel

Charge

10 20 30 40 50 60 500 U bd = 26,06±0,04

Time constant τ, ns x10

3 Nphel

Recovery time

E.Popova SiPM SiPM Nonlinearity... 26 June 14 2018

Pulse shape for different light Intensities. MEPHI data Hamamatsu S10362-11-100U No.50, Ubreakdown=68.4V, U=69.5V

10 15 20 25 30 0,00 0,05 0,10 0,15 0,20 Amplitude, V time, ns

from 0.4 phe/cell to 30k phe/cell

10 15 20 25 30 35 40 45 50

0,00 0,02 0,04 0,06 0,08 0,10 0,12 0,14

relative amplitude, a.u. time, ns Normalized pulse shapes Fast part Slow part (recovering)

) ( *

slow fast quench

C C R + = τ

A=Ntotal_cell*ΔU/Rquench* 50[Ohm] ΔU changes with intensity – potential drops on cell p-n-junction below Ubreakdown Q changes with intensity τ doesn’t change with intensity

E.Popova SiPM SiPM Nonlinearity... 27 June 14 2018

50 100 150 200 250 300 350 2 4 6 8 10

high las er intens ity medium las er intens ity s mall las er intens ity

Q, nVs Ttime, ns

Recovery time for high light Intensities (many phe/cell). Double light pulses method Hamamatsu S10362-11-100U No.50, Ubreakdown=68.4V, U=69.5V

500 1000 4 8

high las er intens ity medium las er intens ity s mall las er intens ity Q, nVs Time Δt, ns

Time shifting t-t0

Q Integration time 50 ns

Fixed intensity Variable intensity Δt As higher intensity of the first pulse as longer time Δt before second pulse starts to give Geiger discharge (1); But recovery constant is the same (2) 1) 2) Breakdown voltage No Geiger Voltage on p-n-junction U t

E.Popova SiPM SiPM Nonlinearity... 28 June 14 2018

Important image! To analyze SiPM waveform one needs to be sure that there are no external network influence Fast component (geiger discharge) Slow component (pixel recovery)

where Ceq= N*[CqCd/(Cq+Cd)]

RL

In case of Rq>>N*RL

N- total number

• f cells in SiPM

E.Popova SiPM SiPM Nonlinearity... 29 June 14 2018

Recovery time depends on number of fired pixels and load resistor

Studying Voltage Recovery Processes on Silicon Photomultipliers Instruments and Experimental Techniques, 2013, Vol. 56, No. 6, pp. 697–705

If

E.Popova SiPM SiPM Nonlinearity... 30 June 14 2018

Experimental study of a SiPM recovery time

Recovery Time of Silicon Photomultiplier with Epitaxial Quenching Resistors Instruments 2017, 1, 5; doi:10.3390/instruments1010005

For 3 x3 mm2 SiPM, with 90 000 pixels the 90000 pixels the recovery time is 31.1 +-1.8 ns; 2000 pixels 6.5 +-0.4 ns

• ne fired pixel 3.1 +- 0.2 ns.

For 1.4 x1.4 mm2 device, ~20 000 pixels 15 000 pixels the recovery time is 15.2 +-0.5 ns

E.Popova SiPM SiPM Nonlinearity... 31 June 14 2018

Summary:

Work has been supported by Megagrant 2013 program of Russia, agreement № 14.А12.31.0006 from 24.06.2013

• SiPM charge, recovery time and amplitude

depend on light intensity;

• Depending on SiPM cell construction (technology used) high light

intensities may affect cell capacitance and/or cause enhanced voltage drop

• n cell pn-junction (below Ubreakdown);

Possible reasons for such behavior:

• conventional feedback between ionization rates and instant pn-junction
• vervoltage becomes too “slow” for extremely fast and strong Geiger

discharge development

• very local feedback due to screening effect of free carriers produced

during ionization in depletion region starts play a role in this case. For Geiger discharge in oversaturated conditions (>>1 phe/SiPM cell)

E.Popova SiPM SiPM Nonlinearity... 32 June 14 2018

E.Popova SiPM SiPM Nonlinearity... 33 June 14 2018

5000 10000 15000 20000

1 2 3 fas t part s low part full charge

U=35V , T =+23

0C

U bd = 33,25±0,05 (T =+23,5±1

0C )

Relative area Nphel

E.Popova SiPM SiPM Nonlinearity... 34 June 14 2018

Hamamatsu MPPC 100U 1x1mm2

“One possible explanation could be that a very high number of input photons per pixel may trigger several avalanches simultaneously, giving rise to a slightly higher output signal compared to the single photon signal.” L. Gruber et al.NIM A737 (2014) 11–18

Relative amplitude

E.Popova SiPM SiPM Nonlinearity... 35 June 14 2018

Not C but Q dI/dV = Ncell·d(ΔQ/ Δ t)/ /dV

) ( * ) (

br slow fast

U U C C Q − + =

rep microcell cell

F C N dU dI * * / =

) (

slow fast microcell

C C C + =

E.Popova SiPM SiPM Nonlinearity... 36 June 14 2018

Over saturation behavior of SiPMs at high photon

• L. Gruber et al./NIM A737 (2014) 11–18

Advanced Laser Diode Systems (PIL040) 404 nm, 20kHz,FWHM 32 ps Relative amplitude Amplitude analysis of 1x1mm2 different SiPMs It has been reported that MPPC pulse shape doesn’t depend on light intensity Used Amp might be the reason for that