beam loss monitoring

Beam Loss Monitoring – Detectors

Photon Detection and Silicon Photomultiplier Technology in accelerator and particle physics

Sergey Vinogradov

QUASAR group

Department of Physics, University of Liverpool, UK

Cockcroft Institute of Accelerator Science and Technology, UK

P.N. Lebedev Physical Institute of the Russian Academy of Sciences, Russia

Sergey Vinogradov oPAC Advanced School on Accelerator Optimization Royal Holloway University, London, UK, July 9th, 2014

Content

◙ 1. Introduction: the best photodetectors

◙ 2. Silicon Photomultipliers (SiPM) as new photon number resolving detectors

◙ 3. Benefits, drawbacks, and typical applications of SiPM

◙ 4. Evaluation studies of SiPMs for Beam Loss Monitoring

◙ 5. Modelling and analysis of comparative performance: SiPM vs PMT and APD

◙ 6. Trends and prospects of SiPM technology for BLM and accelerator applications

Sergey Vinogradov oPAC Advanced School on Accelerator Optimization Royal Holloway University, London, UK, July 9th, 2014 2

Introduction

◙ 1. Introduction: the best photodetectors your choice?







Photodetector #1 *


Adaptive focusing & trichromatic / monochromatic vision High sensitivity due to 100 million rod cells (10-40 photons)High resolution & double dynamic range due to 5 million cone cellsHigh readout rate of 30 frame/sInternal signal processing (100M cells to 1M nerves @30fps)540 million years old design

(*) Yu. Musienko, NDIP 2011

CCD/CMOS approach – toward to #1


Trichromatic / monochromatic vision Number of pixels up to ~ 50 M Sensitivity from ~ 10 -100 photonsDynamic range up to ~ 50KReadout up to ~ 1000 frame/s40 years old design

SiPM approach – toward to ideal low photon detection


R 50

h

pixels

Ubias 50V

Al

DepletionRegion2 m

substrate

Resistor ~1 MOhm

20m

42m

R 50

h

pixels

Ubias 50V

Al

DepletionRegion2 m

substrate

Resistor ~1 MOhm

20m

42m

20th Anniversary ~ now!

Silicon Photomultipliers (SiPM) as new photon number resolving detectors








Concept of ideal detector:first step to SiPM

◙ Ideal detector: conversion of any input signal starting from single photon to recognizable output without noise and distortion in amplitude and timing of the signal

W. Farr, SPIE LEOS 2009

Ideal photon detector

Real photon detector


Single photon detectionin 1 GHz BW with electronic noise 104 e

5 104 0 5 104 1 105

Single photon outputExternal noise

Photodiode

Output, electrons

Pro

babi

lity

dis

trib

utio

n

5 104 0 5 104 1 105


APD

Output, electrons

Pro

babi

lity

dis

trib

utio

n1 105 9.5 105 2 106


PMT

Output, electrons

Pro

babi

lity

dis

trib

utio

n

y 700000 1100000 2000000

1 105 9.5 105 2 106


SiPM

Output, electrons

Pro

babi

lity

dis

trib

utio

n

FactorNoiseExcessN

NENF

N

NResolution

out

out

out

noiseout

2

2

2

22

)(1

)(

σ(Nout)

σnoise

Gain =1ENF ~ 1

Gain ~ 100ENF ~ 3…10

Gain ~ 1 MENF ~ 1.2 Gain ~ 1 M

ENF ~ 1.01


Avalanche with negative feedback: main step to SiPM

Strong negative feedback = fast quenching & small charge fluctuations

0 100 200 3000

1 104

2 104

3 104

0

2 105

4 105

6 105

Avalanch current (high)Avalanch current (low)Electric field (high)Electric field (low)

Avalanche breakdown with negative feedback

Time, ps

Ava

lanc

he c

urre

nt, e

/ps

Inte

rnal

ele

ctri

c fi

eld,

V/c

m

Higher Field

+Current filament

avalanche layer

Accumulatedminority carriers(inversion layer)

feedback layer (wide-bandsemiconductor)

depleted andquasineutral

layers

metallic electrode Initialphotoelectron

p-Si

V. Shubin, D. Shushakov, Avalanche Photodetectors, 2003


Multi-pixel design & feedback resistor:final step to SiPM

Fig. 1-4: Sadygov, NDIP 2005 Fig. 5-7: B. Dolgoshein et al., 2001-2005

R 50

h

pixels

Ubias 50V

Al

DepletionRegion2 m

substrate

Resistor ~1 MOhm

20m

42m

R 50

h

pixels

Ubias 50V

Al

DepletionRegion2 m

substrate

Resistor ~1 MOhm

20m

42m

SiPM – 1996 / 2000sMRS APD – 1990s


SiPM: photon number resolution

B. Kardinal et al., Nat. Photonics, 2008

R. Mirzoyan et al., NDIP, 2008

A. Barlow and J. Schilz, SiPM matching event, CERN, 2011

APD (self-differencing mode) VLPC SiPM (MEPhI/Pulsar)

SiPM (Excelitas)PMT (Hamamatsu R5600)

I. Chirikov-Zorin et al, NIMA 2001

MPPC (Hamamatsu)

S. Vinogradov, SPIE 2011


Sergey Vinogradov Seminars on SiPM at the Cockcroft Institute 2 December 2013 13G. Collazuol, PhotoDet, 2012

Benefits, drawbacks, and typical applications of SiPM








SiPM: photon number resolution

◙ SiPM looks like ~ ideal detectorNear-ideal amplification:

― Gain > 105, ENF < 1.01

Room temperature

Low bias (<100 V)

Large area (6x6 mm2)

Good timing (jitter < 200 ps)

Fast response (rise <1 ns, fall~20 ns)

◙ In fact, not a photon spectrumPhotoelectrons

Dark electrons

Crosstalk & Afterpulses

◙ In fact, non-Poissonian distributionWhy?

How much?

Distribution?

Resolution? P. Finocchiaro et al., IEEE TNS, 2009

A. Barlow and J. Schilz, SiPM matching event, CERN, 2011

SiPM (Excelitas)

2

][1][

X

XVarXENF


SiPM drawbacks: crosstalk

◙ Crosstalk: hot carrier photon emission + detection = false event

A. Lacaita et al., IEEE TED, 1993

R. Mirzoyan, NDIP, 2008

Yu. Musienko, NDIP, 2005


SiPM drawbacks: afterpulsing

◙ Afterpulsing: trapping + detrapping + detection = false event

primary avalanche

afterpulses

Δ time

Output

G. Collazuol, PhotoDet, 2012 C. Piemonte et al., Perugia, 2007


SiPM drawbacks: nonlinearity

◙ Limited number of pixels = losses of photons◙ Dead time of pixels during recovery = losses of photons

Plot details:

Npixel=100

PDE=100%

No Noise (DCR, CT, AP)

0

20

40

60

80

100

0 50 100 150 200 250 300

Number of photons, Nph

Nu

mb

er o

f fi

red

pix

els,

Ns

Ideal response<Ns>=<Nph>

Nonlinear response<Ns>=f(<Nph>)

Output SignalDistribution

Equivalent SignalDistribution

Incident PhotonDistribution

( )

1eq Nph Ns d Ns

d Nph

Ideal 100 pixel SSPM

Idea

l pho

ton

dete

ctor


Application types

◙ Binary detection of light pulses – “events” (bit error rate) - NA

◙ Photon number resolution (noise-to-signal ratio, σn/μn) - Calorimetry

◙ Time-of-flight detection (transit time spread, σt) – TOF PET

◙ Detection of arbitrary signals starting from photon counting - Iph(t)

- Beam Loss Monitoring


SiPM application examples

◙ CalorimetrySiPM (MEPhI) small HCAL (MINICAL), DESY, 2003

MPPC (Hamamatsu), T2K, 2005-2009

MPPC at LHC CMS HCAL

RICH for ALICE (LHC)

FermiLab, Jefferson Lab calorimeter upgrade projects

◙ AstrophysicsSiPM cosmic ray detection in space (MEPhI, 2005)

Cherenkov light detection of air showers (CTA, 2013)

◙ Medical imaging Positron Emission Tomography:

―TOF-PET ― PET / MRI

◙ TelecommunicationQuantum cryptography

Deep space laser link (Mars exploring program)


Evaluation studies of SiPMs for Beam Loss Monitoring








Beam Loss Monitoring (ref. E. Nebot talk 08-07-14)

Objectives:

ProtectMonitorAdjust


BLM: first evaluation of SiPM

D. Di Giovenale et al., NIMA, 2011SPARC accelerator, Frascati, INFN

FERMI@Elettra, Synchrotrone Trieste

MPPC, 1mm2, 400 pixels

Quartz fiber 300 μm, 100 m

Dark count noise: negligible

Electronic noise: negligible

Spectral dispersion in fiber: n( ) →∆t( ) 𝜆 𝜆 ~ 3 ns @100 m

τfall ~ 10 ns → deconvolution◙Compact low cost BLM

1m-scale resolution @100 m


SiPM performance metrics for BLM

◙ Loss scenarios reconstructionAmplitude → # photons → # particles per location (PNR)

Transit time to rising edge → single loss location (Time Res.)

Resolution of multiple loss locations & # particles ― Modulation transfer function (MTF) ?― Nonlinearity has to be accounted !

PNR

TTS

New metrics ?


Challenges for SiPM in BLM:saturation, recovering, duplications

◙ Transient nonlinearity of SiPM responseLarge rectangular light pulse: Nph > Npix; Tpulse > Trec

Peak – initial avalanche events in ready-to-triggering pixels

Plateau – repetitive recovering and re-triggering of pixels

Fall – final recovering (without photons, but with afterpulses!)

4 us pulse 50 ns pulse


MPPC response on rectangular pulse


Modelling and analysis of comparative performance: SiPM vs PMT and APD








Photon Number Resolution

◙ Photon Number Resolution & Excess Noise FactorBurgess variance theorem

0

20

40

60

80

100

0 50 100 150 200 250 300


Nu

mb

er o

f fi

red

pix

els,

Ns

Ideal response<Ns>=<Nph>

Nonlinear response<Ns>=f(<Nph>)

Output SignalDistribution

Equivalent SignalDistribution

Incident PhotonDistribution

Ideal 100 pixel SSPM

Idea

l pho

ton

dete

ctor

1

2

2

2

2

2

2

1

)()(

)(

)(

1)(

1

ENFENFPoissonianNif

ENFN

N

N

NPNR

randomNif

N

NN

N

ENF

NifN

NENF

Nin

Nin

in

out

outout

in

in

in

out

out

N

in

out

out


Schematics of AP & CT stochastic processes

Branching Poisson

Crosstalk Process

Geometric Chain

Afterpulsing

Process

Poisson number of primaries <N>=μ

e.g. SSPM Photon Spectrum

Single primary event N≡1

e.g. SSPM Dark SpectrumDuplication

Models

Non-random (Dark) event

Primary

1st CT

2nd CT

No CT

Ran

dom

CT

eve

nts …

Random CT events

Random Photo eventsRandom primary (Photo) events

Ran

dom

CT

eve

nts

…

…

…

Non-random (Dark) event

Ran

dom

CT

eve

nts …

… …

…

…

…

Random primary (Photo) events

Ran

dom

CT

eve

nts

…

…


Results Geometric chain process Branching Poisson process

Primary event distribution

Non-random single (N≡1)

Poisson (μ) Non-random single (N≡1)

Poisson (μ)

Total event distribution

Geometric (p) Compound

Poisson (μ, p) Borel (λ) Generalized Poisson (μ, λ)

P(X=k) pp k 11

Ref. [1]

!

exp1

k

kk k

!

exp1

k

kk k

E[X] p1

1

p1

1

1

1

Var[X] 2)1( p

p

2)1(

)1(

p

p

31

31

ENF p1 )...(2

31

1

1 32 ppp

Analytical results for CT & AP statisticsCT & AP model results

[1] S. Vinogradov et al., NSS/MIC 2009 λ is a mean number of successors in one branch generation


1.0E+01

1.0E+02

1.0E+03

1.0E+04

1.0E+05

1.0E+06

1.0E+07

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18

Threshould [fired pixels]

Dar

k C

ou

nt

Rat

e [H

z]

Experiment

Borel

Geometric

A few photon detection spectrum of Hamamatsu MPPC

(S. Vinogradov, SPIE 2012)

Dark event complimentary cumulative distribution – DCR vs. threshold

(S. Vinogradov, NDIP 2011; experiment B. Dolgoshein et al., NDIP 2008)

Pct=40%

Pct=10%

P. Finocchiaro et al., IEEE TNS, 2009

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

-1 0 1 2 3 4 5 6 7 8 9 10 11 12 13

Number of fired pixels

Pro

bab

ility

de

nsity

fun

ctio

n

Experiment

Generalized Poisson

A.N. Otte, JINST 2007

Crosstalk models and experiments


<Ns>=f(<Nph>)

0 50 100


SSPM

sign

al, fi

red

pixe

ls Ns

0

50

Photon signaldistributionSSPM equivalentphoton distributionCalibration

SSPM signaldistribution

NphdNsdNsNph

1

cell

ph

N

PDEN

N

PDEN

N

PDEN

cellNN

PDEN

cellfired

N

PDENe

ENF

eeNeNN

cell

ph

cell

ph

cell

ph

fired

cell

ph

1

11 2

SiPM binomial nonlinearity

Intrinsic Resolution in σ units; Npix=506

Photons per pulse, Nph

B. Dolgoshein et al., 2002Ou

tpu

t si

gn

al, N

s

E.B. Johnson, NSS/MIC 2008


SiPM recovery nonlinearity

1 10 100 1 103 1 104 1 1050.01

0.1

1

Tpulse=20TresetTpulse=2TresetTpulse<<Treset

Dead-time non-linear Resolution

Number of photons per pulse

1

0.01

R Nph 5 10 9 R Nph 50 10 9 R Nph 500 10 9

1 1051 Nph

Plot details:

PDE=100%

Npixel=500

Tpulse=100 ns

32

)1(

1

dead

Ns

dead

Ns

t

t

Nonparalizible dead time modelProbability distribution (~ Gaussian)

W. Feller, An Introduction to Probability Theory and Its Applications, Vol. 2, Ch. XI, John Willey & Sons, Inc., 1968

( ) ( ) 1eq Nph deadENF

Recovery non-linearity → ENFS. Vinogradov et al., IEEE NSS/MIC 2009

pulse recT T

2) Exponential recovery of Gain m(t) accounting for Pdet(m)S. Vinogradov, SPIE DSS, 2012

3

( ) det

1( ) ( )

12

receq Nph

rec

ENF if P m const

M.Grodzicka, NSS 2011


Performance metrics: ENF and DQE

Noise source Key parameter Typical values for

SiPM ENF expression

SiPM ENF

PMT ENF

Fluctuation of multiplication

Mean gain and standard deviation of gain

μ(gain) ~ 106

σ(gain) ~ 105 )(

)(1

2

2

gain

gainFm

1.01 1.2

Crosstalk Probability of crosstalk

event Pct 5% – 40%

1

1 ln(1 )Fct

Pct

1.05 – 2 1

Afterpulsing Probability of afterpulse

Pap 5% – 20% 1Fap Pap 1.05 – 1.2 1.01

Shot noise of dark counts

Dark count rate DCR Signal events in Tpulse

105 – 107 cps 1DCR Tpulse

FdcrNph PDE

1.01 1

Noise / losses of photon detections

Photon detection efficiency PDE

15% – 40% 1

FpdePDE

2.5 – 6.6 5

Total noise in linear dynamic range

Total ENF_linear _ENF linear Fm Fct Fap Fdcr Fpde 2.8 – 16 6

Pixel number nonlinearity

Mean number of possible single pixel firings n

exp( ) 1; ; binomial distribution

Nph PDE nn ENFpix

Npix n

Recovery time nonlinearity

Mean rate of possible single pixel firings λ

; 1 ; nonparalizible modeln

ENFrec TrecTpulse

totalnlapctdcrpdemtotaltotalin

out

outout ENF

DQEFFFFFFENFNphENFNphPNRN

NPNR

1)()(

)(

)(

*

*


Performance in DQE – various detectors

SNR performance (relative to ideal) in nanosecond light pulse detection by 1mm2 PDs

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

100.00%

1.E+0 1.E+1 1.E+2 1.E+3 1.E+4 1.E+5 1.E+6 1.E+7 1.E+8 1.E+9


Det

ecti

ve Q

uan

tum

Eff

icie

ncy

SSPM potential

DAPD Amplification Tech.

MPPC-50 Hamamatsu

APD

PIN - photodiode

PMT

Fm

PDE

Gain

Gain


Time resolution

◙ Time resolution is combined as a sum of contributionsTransit time spread of photon arrival, avalanche triggering, avalanche development, and single electron response times

Jitter of signal amplitude fluctuations in a time scale


Filtered point process approach to amplitude fluctuations & time resolution

Point Process

Marked Point Process

Filtered Output

Clastered Point Process


Clastered filtered point process model

◙ Time resolution includes all essential factors and combines performance in time response and PNR (ENF)

2

2

sec

2

22

2

sec

2sec

secsec

sec

sec

2sec

2

22

sec222

secsec

1

)()(:

)()(

)(1)(

)(

)(

)()(

Thresh

noisetotal

phriset

sersersptrph

out

noisesci

sersptrph

sersptrphgain

pesersptrph

sersptrph

sersptrphpe

ser

noisesersptrphpescisersptrphgainpe

t

V

VENF

N

ththslowisSERandfastispulselightIf

tV

V

th

thENFENF

N

dt

thd

th

dt

thdN

KVV

thNthENFENFN


Time resolution:scintillation model & experiment

asdasd

0 5 10 15

0.14

0.16

0.18

0.2Poisson with SER filtering (no noise)+ Gain and electronic noise+ Scintillator resolution 9%+ Crosstalk 15%+ Crosstalk 30%

Trigger threshold (SER amplitudes)

CR

T, n

s

Most popular & demanded case study: LYSO+MPPCLYSO: 0.09 ns rise, 44 ns decay; 9% resolution

MPPC: Npe=3900, ENFgain=1.015, Pct=0.14; SPTR=0.124 ns, Vnoise=0.32 mV S. Seifert et al, “A Comprehensive Mode to Predict the Timing Resolution ”, TNS, 2012.

MPPC SER pulse shape – analytical expression (~ 1 ns rise, ~ 25 ns decay) D. Marano et al, “Silicon Photomultipliers Electrical Model: Extensive Analytical Analysis” TNS 2014


Arbitrary signal detection:rectangular pulse response model


Trends and prospects of SiPMs for BLM and accelerator applications








SiPM trends and advances

◙ Market leadersHamamatsuKETEKSensLFBK / AdvanSiDExcelitas / Perkin ElmerPhilips (digital SiPM)

◙ Design improvements (~ in a few year time scale)Higher Photon Detection Efficiency (30% → 70%)Lower crosstalk, lower afterpulsing (30% → 3%)Lower dark count rate (1000 → 40 Kcps/mm2)Faster SER, smaller pixel size (25 → 10 um)Larger area, larger arrays (3x3→10x10mm2, 4x4 → 16x16 channels)

Latest news from 2nd SiPM Advanced Workshop and Conf . on New Development s in Photodetection, 2014


Hamamatsu: Through Silicon Vias


Hamamatsu: Low Crosstalk & Afterpulsing


Performance in DQE - MPPC series

Pulse duration & detection time = 10 ns

1 103 0.01 0.1 1 10 100 1 10

3 1 104 1 10

5 1 106 1 10

70

0.1

0.2

0.3

0.4

MPPC 100 umMPPC 50 umMPPC 25 umMPPC 15 umMPPC 10 um

DQE of market MPPCs (3x3mm 100, 50, 25, 15, 10 um pixels)

Photons / pulse

DQ

E

1 103 0.01 0.1 1 10 100 1 10

3 1 104 1 10

5 1 106 1 10

70

0.1

0.2

0.3

0.4

0.5

MPPC 100 um trenchMPPC 75 um trenchMPPC 50 um trench

DQE of newest MPPCs (3x3 mm 100, 75, 50, um pixels)

Photons / pulseD

QE


KETEK

◙ Highest PDE @50 um pixels◙ Various geometries◙ 15 … 100 um pixels


SensL

◙ Fast capacitive outputFWHM < 3.2 ns @ 6x6 mm2

◙ Large arrays / modules◙ Low cost


Philips: Digital SiPM (Modern active quenching SPAD array)


Summary on BLM with SiPM

◙ BLM is one of the most challenging application for SiPMBenefits

―Practical & efficient (cost, compactness, Si reliability…)―Perfect Transit Time Resolution (as is for now)―Acceptable DQE within dynamic range (may be better)

Drawbacks―Upper margin of dynamic range is low (design improvement)

– Number / density of pixels (↑ 10 times)– Pixel recovery time (↓ 10 times)

―Time response (bandwidth) (external measures)– Analog / digital SiPM output signal processing

◙ BLM with SiPM: big problem with a chance to winAnd with a lot of space for new ideas, designs, and fun

Sergey Vinogradov Seminars on SiPM at the Cockcroft Institute 3 February 2014 49

SNR performance (relative to ideal) in nanosecond light pulse detection by 1mm2 PDs

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

80.00%

90.00%

100.00%

1.E+0 1.E+1 1.E+2 1.E+3 1.E+4 1.E+5 1.E+6 1.E+7 1.E+8 1.E+9


Det

ecti

ve Q

uan

tum

Eff

icie

ncy

SSPM potential

DAPD Amplification Tech.

MPPC-50 Hamamatsu

APD

PIN - photodiode

PMT

Summary on SiPM

◙ SiPM technology: breakthrough in photon detectionPhoton number resolution at room temperature

Silicon technology / mass production / reliability / price

Highly competitive in short (< μs) pulse detection

Fast progress in improvements: DQE, Dynamic range, Timing

◙ Welcome to

SiPM applicationsScintillation

Cherenkov

Laser pulse

And much more…


The end

◙ Thank you for your attention

◙ Questions?

[email protected]


mailto:[email protected]

beam loss monitoring – detectors photon detection and silicon photomultiplier technology in...

Documents

silicon photomultipliers

typical applications

prospects of sipm technology

sipm vs pmt

new photon number

university of liverpool

medical applications