Large pixel SiPMs for single photon detection in the new LHCb large - - PowerPoint PPT Presentation

large pixel sipms for single photon detection in the new
SMART_READER_LITE
LIVE PREVIEW

Large pixel SiPMs for single photon detection in the new LHCb large - - PowerPoint PPT Presentation

PIXEL2018 International workshop on Semiconductor Pixel Detectors for Particles and Imaging Large pixel SiPMs for single photon detection in the new LHCb large area scintillating fibre tracker Olivier Girard , Maria Elena Stramaglia, Guido


slide-1
SLIDE 1

Olivier Girard, Maria Elena Stramaglia, Guido Haefeli

Ecole polytechnique fédérale de Lausanne (EPFL), Switzerland On behalf of LHCb SciFi collaboration

PIXEL2018 International workshop on Semiconductor Pixel Detectors for Particles and Imaging

Large pixel SiPMs for single photon detection in the new LHCb large area scintillating fibre tracker

slide-2
SLIDE 2

Outline

Context

  • LHCb and the upgrade
  • The SciFi technology

Silicon photomultipliers

  • Choices for LHCb

Characterisation of radiation effects

  • Random noise
  • Detection of low photon signals
  • Photon detection efficiency
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

2

  • B. Leverington

Module assembly on frame

slide-3
SLIDE 3
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

3 3 Tracking station 12 detection layers 320 m2 total area 2× 3 m 2× 2.5 m SciFi module fibre mats + mirrors Read-out box SiPMs, cooling, electronics mirrors

LHCb SciFi tracker

✘ Straw tubes + silicon? ✓ SciFi tracker

slide-4
SLIDE 4
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

3 3 Tracking station 12 detection layers 320 m2 total area 2× 3 m 2× 2.5 m SciFi module fibre mats + mirrors Read-out box SiPMs, cooling, electronics mirrors

LHCb SciFi tracker

✘ Straw tubes + silicon? ✓ SciFi tracker

slide-5
SLIDE 5
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

3 3 Tracking station 12 detection layers 320 m2 total area 2× 3 m 2× 2.5 m SciFi module fibre mats + mirrors Read-out box SiPMs, cooling, electronics mirrors

LHCb SciFi tracker

✘ Straw tubes + silicon? ✓ SciFi tracker

  • Fibres: 11’000 km
  • SiPMs: 0.2 m2, 490k channels
slide-6
SLIDE 6
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

4

Light detection – challenge

Signal =

Injection at mirror DV=3.5V

X = 1.35 mm X/X0 = 0.41% Energy deposit × scint. light yield × light capture × light attenuation × photon det. eff.

VELO upgrade

Signal = O(10k) e- Noise = O(200) e-

Collected

  • n 1 pixel

X = 220 μm X/X0 = 0.23%

slide-7
SLIDE 7
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

4

Light detection – challenge

Signal =

Injection at mirror DV=3.5V

X = 1.35 mm X/X0 = 0.41% Energy deposit × scint. light yield × light capture × light attenuation × photon det. eff. → Amplification needed!

VELO upgrade

Signal = O(10k) e- Noise = O(200) e-

Collected

  • n 1 pixel

X = 220 μm X/X0 = 0.23%

slide-8
SLIDE 8
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

4

Light detection – challenge

Signal =

Injection at mirror DV=3.5V

X = 1.35 mm X/X0 = 0.41% Energy deposit × scint. light yield × light capture × light attenuation × photon det. eff. → Amplification needed! Noise Collected from 2-3 ch. = 2-3 × 104 pixels

VELO upgrade

Signal = O(10k) e- Noise = O(200) e-

Collected

  • n 1 pixel

X = 220 μm X/X0 = 0.23%

slide-9
SLIDE 9
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

4

Light detection – challenge

Signal =

Injection at mirror DV=3.5V

X = 1.35 mm X/X0 = 0.41% Energy deposit × scint. light yield × light capture × light attenuation × photon det. eff. → Amplification needed! Noise Collected from 2-3 ch. = 2-3 × 104 pixels Radiation environment Fibres: max 30kGy SiPMs: 5·1011 neq/cm2

VELO upgrade

Signal = O(10k) e- Noise = O(200) e-

Collected

  • n 1 pixel

X = 220 μm X/X0 = 0.23%

slide-10
SLIDE 10

SiPM for the SciFi tracker

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

5

Customised SiPM array produced by Hamamatsu (S13552)

Design choices:

 Large pixels  Opaque trenches  Adjusted quenching resistor

32.54 × 1.625 mm2

slide-11
SLIDE 11

Dark count rate and effect of irradiation

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

6

Random noise – Dark count rate (DCR)

  • Thermally-generated e-h pair ~100 kHz/mm2
  • Increased by irradiation 30’000×

 Overlap of pulses (high signals)  Cooling (÷2 per 10°C)  Short integration time  Clustering

0.41mm2 channel DV=3.5V T=-40°C 4.5V 3.5V 2.5V neutrons protons

slide-12
SLIDE 12

Sensitivity for low-light signals after irradiation

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

7

DV=2V T=-40°C

  • 11%
  • 7%
  • 4%

SPIROC

Low light spectrum

  • Gain measurement
  • Compensation for effects in the electronics

due to the high dark current

PACIFIC

DV=3.5V T=-40°C 6∙1011 neq/cm2 Non-irradiated

slide-13
SLIDE 13

Light yield with irradiated SiPMs

Measurement

  • Correction for DCR contribution
  • 5% measurement uncertainty
  • Light yield variation observed:
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

8

“Light yield” Noise ±4% up to 6∙1011 neq/cm2

Electron injection in the fibres

  • Comparative photon detection

efficiency

DV=3.5V T=-40°C

SPIROC

slide-14
SLIDE 14

Conclusion & outlook

SciFi technology

  • Cost-efficient solution to cover large surface with good spatial

resolution (<100 μm for 2.5 m, better for small size)

  • Particularly interesting for applications with no radiations
  • Can provide timing information ~0.5 ns

How can we make the SciFi technology radiation-harder?

Photodetector:

  • SiPMs: improve structure and implementation
  • Use a pixel sensor with amplification for visible light detection?
  • Improve light collection: microlenses

Fibre:

  • Improve scintillation light yield in the green, reduce attenuation and

improve radiation hardness

  • Light transport outside the radiation environment with clear interface

with optical fibre?

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

9

slide-15
SLIDE 15

Backup

LHCb & the SciFi tracker

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

15

slide-16
SLIDE 16

LHCb upgrade

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

16

Timescale during LS2 in 2019-2020 Goal extend physics reach with:

 Higher luminosity (5×)  Better trigger efficiency for a wide

range of decay channels

 Design for 50 fb-1 integrated

luminosity Run1+2 (2011-now): 7 fb-1 Detector changes

 40 MHz read-out + flexible trigger (in

software)

 Detector hardware: cope with

increased occupancy and read-out rate Downstream tracker is replaced: silicon strips (IT) + straw tubes (OT) → scintillating fibre (SciFi) tracker modules

 High hit detection efficiency  Fine granularity  Low mass (homogenous distribution)

slide-17
SLIDE 17

SciFi tracker

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

17 2× 3 m 2× 2.5 m

  • Total area of 320 m2
  • 11’000 km fibres (Ø250

µm) arranged in mats

  • 4000 multi-channel

Silicon PhotoMultipliers (SiPMs) for a total of 524k read-out channels

6 layers 1.35 mm

Tracking station 4 detection layers (stereo angles 0, ±5°) SciFi module fibre mats + mirrors Read-out box SiPMs, cooling, electronics mirror s

30 kGy 5·1011 neq/cm2

slide-18
SLIDE 18

Simulation of the cumulated radiation after the design integrated luminosity (50 fb-1)

Radiation environment

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

18

Fibres:

 In the central region: 30 kGy

SiPMs:

 Dominated by neutrons  Neutron shielding between SciFi tracker

and calorimeter

 Fluence expected: 5·1011 neq/cm2

Effect on the performance:

 Reduced transparency of the fibres  Increase in SiPM noise

M1 replaced by polyethylene with 5% boron (10-20cm thick) Calorimeter SiPMs

SciFi tracker Neutron fluence: ratio

"𝑥𝑗𝑢𝑖 𝑡𝑖𝑗𝑓𝑚𝑒𝑗𝑜𝑕" "𝑜𝑝 𝑡𝑖𝑗𝑓𝑚𝑒𝑗𝑜𝑕"

30 kGy 50 to 100 Gy + neutrons

SiPMs

Total ionising dose at the end of the lifetime of the SciFi tracker [Gy]

SiPMs F i b r e s

slide-19
SLIDE 19

SciFi tracker

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

19

Challenges and design choices:

 Signal time spread: scintillator

decay time and light transport

 Attenuation length: ~ size  Large size: flatness and

mechanical stability with low material budget

 Radiation environment (fibres and

SiPMs): detection efficiency

 SiPM cooling (-40°C)

Kuraray SCSF-78MJ blue emitting fibre

Cold box cross-section

SiP M 3D-printed Ti cooling pipe Cold-warm feed-through FE electronics connection Cold box isolation Fibres 3D-printed Ti cooling pipe Enclosure Vacuum insulated distribution pipes

slide-20
SLIDE 20

 SiPM gluing to cooling pipe and optical

alignment

 3D printed Ti cooling pipe with alignment

pins, thermal expansion and isolation are the main challenges

 Integration into a vapour tight cooling

enclosure, vacuum insulated cooling pipes

 Cooling with single phase liquid chiller

(Novec or C6F14)

 Front-end electronics with custom read-out

chip (Pacific)

 Zero suppression (clusterisation) based on 3

threshold sampling

 Optical transmission, zero suppression on

FPGA and GBT transmission

 Common off detector electronics TELL40

Integration of SiPMs in LHCb SciFi

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

20 Multi-channel SiPM with feed-through Cooling pipe Enclosure Vacuum insula- ted connections

Cold- box

slide-21
SLIDE 21
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

21

IEEE NSS MIC 2017

slide-22
SLIDE 22

Backup

SiPM

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

22

slide-23
SLIDE 23

Silicon photomultipliers – working principle

Based on avalanche photodiodes in GM-mode

  • Internal amplification (gain ~106) through impact ionisation
  • Operated above the breakdown voltage VBD (GM-APD)
  • Quenching circuit (ex. RQ) and cell recovery
  • Photon counting: array of GM-APDs in parallel
  • Photon detection efficiency
  • PDE = QE(l) × FF × P01(DV)
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

23 O(1 - 10 mm) O(10 - 100 µm)

HPK, thin metal film RQ KETEK, Custom implementation HPK, MPPC

slide-24
SLIDE 24

Noise sources

Random noise – Dark count rate (DCR)

  • Thermally-generated e-h pair ~100 kHz/mm2
  • Increased by irradiation 30’000×

Overlap of pulses (high signals) Reduced by cooling (÷2 per 10°C) and

short integration time window

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

24

Acerbi, 2015

Correlated noise

 Avalanches generated ~simultaneously  DiXT, DeXT, AP  Limits the operation range (and

therefore PDE)

slide-25
SLIDE 25

Working principle bonus

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

25

slide-26
SLIDE 26

Photon detection efficiency

  • Calibration using a photodiode
  • Two independent methods: SiPM

current and pulse frequency under illumination

  • Accurate corrections for correlated

noise (and dark)

  • Estimated uncertainty on PDE: 6.0% /

3.5%

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

26

PDE ∆𝑊 = 𝑐 ∙ 1 − 𝑓−∆𝑊/𝑏 𝑏 𝜇 > 𝑏(𝜇)

QE∙FF P01 For p-on-n structure:

SCSF-78MJ emission spectrum

~50% @ peak

slide-27
SLIDE 27

SiPM electrical model

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

27

𝜐d = 𝑆d ∙ 𝐷d + 𝐷Q 𝜐short = 𝑆load ∙ 𝐷tot 𝜐long = 𝑆Q ∙ 𝐷d + 𝐷Q

slide-28
SLIDE 28

Backup

SiPM characterisation

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

28

slide-29
SLIDE 29

Measurement setups

Single-channel based setups

  • Fast preamplifier and oscilloscope
  • IV scan (ev. with multiplexer)

Multichannel arrays

  • Use integrated front-end ASICs
  • Signal is integrated and/or shaped
  • Pulsed light injection
  • Fibre mat and electron-gun
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

29

Pulse shape Correlated noise RQ DCR Gain PDE VBD VBD Gain PDE Gain NCR

Suitable for irradiated SiPMs (up to 1012 1MeV neq/cm2)

slide-30
SLIDE 30

Oscilloscope setup

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

30

Oscilloscope

  • t-dependent V pulses
  • Statistical functions
  • Remote control
  • Offline analysis

Pulse shape Amplifier Light injection Cooling

  • Pulse shape
  • Correlated noise
  • Irradiated SiPMs
  • Relative PDE

Measurements on single channel Light source Monochromator Fibre Diffuser Calibrated photodiode SiPM PDE

slide-31
SLIDE 31

Pulse shape and correlated noise

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

31

DiXT DeXT AP VBD

slide-32
SLIDE 32

Correlated noise – H2017

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

32

H2017 lot1 DV = 4.8V H2017 lot1 Delayed pulses

slide-33
SLIDE 33

Waveform analysis – irradiated detectors

DiXT Relative gain

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

33

𝑓−𝑢∙𝑔DCR 𝐻 = 𝑉𝑒𝑢 1𝑄𝐹 𝑆load ∙ 𝐻Amp ∙ 𝑓 ~ -6% High DCR DiXT detection

slide-34
SLIDE 34

Noise cluster rate

  • NCR computed for 128-channel

array read out at 40MHz

  • Strong dependence on the

electronics shaping (here: SPIROC)

  • Contributions from random pulse
  • verlap and correlated noise
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

34

Pedestal

DV = 3.5V DV = 3.5V

Dark pulses Correlated noise

Seed thrs = 2.5PE

slide-35
SLIDE 35

Noise cluster rate – simulation

  • Long waveforms acquired on the
  • scilloscope
  • Numerical integration by sections of

length tint

  • Simulation of 128-channel array
  • Clustering: measurement of NCR as a

function of tint

  • Values in the range expected from

measurement with PACIFIC

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

35

Dark spectrum NCR vs seed NCR vs tint 𝜐int

slide-36
SLIDE 36

Dark count rate after irradiation

  • Fast, simple and robust
  • VBD: 𝐽 𝑊

bias = 𝛽 ∙ 𝑊 bias − 𝑊 BD 𝜁

⇒ d ln 𝐽 d 𝑊

bias −1

∝ Δ𝑊

  • RQ(T) for scan in forward bias

region

  • DCR(DV, T)
  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

36

slide-37
SLIDE 37

Backup

SciFi tracker fabrication

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

37

slide-38
SLIDE 38

Cooling

Fibre tracker production

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

38 Custom machine High quality mat: 2.5 m × 13 cm, 6 fibre layers Total of 1200 mats Four production centres (including EPFL) Optical scanner Uniform light yield (β- source) Special modules with beam pipe cut out 5 m × 53 cm Rigidity: honeycomb panels Cooling distribution Neutron shield Cabling

Related posters:

  • P. Hopchev, Production and quality

assurance of scintillating fibre modules for the LHCb upgrade.

Optical properties Mechanical specifications

slide-39
SLIDE 39

Mat production steps

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

39

  • 1. Winding
  • 2. Glue curing with wheel rotating for 12h in

an area with controlled humidity and temperature

  • 4. Foil lamination with

black 25 μm thick capton foil (both sides, ensures light tightness) and end piece glueing

  • 3. Cutting

and unforming

  • 5. Optical cut with a

milling machine (polishing with a diamond head)

  • 6. Mirror

glueing

  • 7. QA tests with optical scanner

and β-source (light yield homogeneity)

  • 8. Delivery to

Heidelberg and Nikhef for module assembly + integration of SiPMs, cooling and FE electronics

slide-40
SLIDE 40

Fibres are produced by Kuraray (300 km every two weeks) and delivered at CERN where quality assurance tests are performed. Goal of QA:

QA for Scintillating fibres

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

40

  • 1. Acceptance test regarding optical and scintillation

properties:

  • Attenuation length with UV LEDs
  • Scintillation yield with beta source
  • Irradiation tests with X-ray source
  • 2. Removal of fibre sections out of mechanical

specification ensuring high quality fibre mat production

  • Fibre defects (large diameter fluctuation) «bump»

identified with laser micrometer scanner

  • Bump shrinking with heating element or cutting

Detection and measurement of defects Fibre spool 12.5 km Target spool Bump shrinking by heating 100°C Ø 350 µm F ≈ 1 N Example: Before 415 µm → After 337 µm

Fiber Bumps larger than 500 µm must be cut away and the fibre re-glued (~15 min, 1-2×/spool). Fibre scanner: 3h30/spool

Bump shrinking is fully automatic and it preserves fibre strength, cladding and 85% light transmission.

slide-41
SLIDE 41

 Four production centres: RWTH Aachen and TU Dortmund (DE), EPFL

(CH) and Kurchatov Institute (RU)

 Custom winding machine produced by an industrial company (one per

winding centre)

 Fibre mat of 2.5 m length × 13 cm width, 6 fibre layers with a total of 7

km of fibres

 Mat winding takes 4h (1 per day)

Fibre mat winding

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

41 Visual monitoring to detect fibre jumps Alignment pin groove in the wheel, filled with glue during winding, allows precision positioning at later production steps ~1200 mats required for the SciFi tracker Aimed production rate: 4 mats/week/site

Fibre spool Threaded Winding wheel Ø 82 cm Fibre tensioning system Linear axis for precision positioning

slide-42
SLIDE 42

Mat transverse geometry:

 Checks for distorted or missing

fibres with an automated optical scanning and fibre detection

Mat response uniformity:

 Measurement of light yield with an

electron source

Mat quality controls

  • O. Girard – PIXEL2018, Academia Sinica, Taipei, Taïwan – 14 December 2018

42 Casted fibre mat End piece Scanner SiPM die gap, partially recovered by neighbouring channels