1

Using MPPCs for T2K Fine Grain Detector

Fabrice Retière (TRIUMF)for the FGD group

University of British Columbia, Kyoto University,University of Regina,TRIUMF and University of Victoria

2

T2K Fine Grain DetectorElement of T2K near detectorActive target for neutrino interactionElements• Plastic scintillator bar

(POPOP)2 meter long

• Light collection with Wavelength Shifting fiber

• Readout by HamamatsuMPPC

• ~10,000 channels

µ

p

1 mm Y11 fiber

0.96 mm

MPPC

ν

3

FGD physics requirements100% efficiency for MIP crossing a barParticle identification• By dE/dx for particle crossing the FGD• By range, especially for stopping protons

Large energy released (10 MIPs)• By detecting Michel positrons for stopping π+

Position resolution• Bar width & no information along the bar

Timing resolution• ~ 3ns per neutrino interaction for matching with

photons in calorimeter

4

MPPC basic parameters

Gain > 2 105

• i.e. 1PE = 2 105 e-Way above typical electronics noise

Photo-detection efficiency• Comparable or better than

PMTBut need to measure PDE for proper wavelength

S. Gomi et al. (Kyoto University)

Delta V [V]0.5 1 1.5 2 2.5

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

PDE Delta V : 400pixel No.050/100

Delta V [V]0.5 1 1.5 2 2.5200

400

600

800

1000

1200

1400

310× Gain Delta V : 400pixel No.050/100

15C20C25C

PDE relative to PMT

69.5V

69.5V

Cross-talk andAfter-pulsing free

5

Photo-electron per MIPMPPC fulfill requirements

Beam test at TRIUMF• 120 MeV/c particles

Electrons are minimum ionizingWorst case scenario• No fiber mirroring• End of the bar

More than 10 direct PE even at 69.5V• No need to run at higher voltage

Issue of Fiber-MPPC coupling still being addressed• New coupler• 1.2x1.2 mm2 MPPC

M. Bryant (UBC), P.Kitching (TRIUMF), S. Yen (TRIUMF)

6

MPPC fulfilling requirementsQuantum efficiency• For 100% efficiency need more than 10 PE per MIP

Go for at least 15 PEs per MIPEnergy resolution. Not directly a MPPC issue• Driven by photon statistics (~25% for 15 PE)

Increase quantum efficiency would helpTiming resolution. • Not a MPPC issue in principle (fast)

Dynamic range• 400 pixels provide more than 50 MIPs dynamic range

due to saturationNuisances: Dark noise, cross-talk, after-pulsing

7

Reading out MPPCs

Compromise between timing resolution and integration time• Desirable to measure all pulses continuously

during beam spill (5 µs) and about 2 muondecay constant (2.2 µs) after spill

• Chose a waveform digitization solutionUse the Switch Capacitor Array designed for Time Projection Chamber (AFTER ASIC)Fairly slow shaper (100 ns rise time)50 MHz sampling frequency512 time bin ~ 10 µs total integration time

8

Waveforms from MPPC coupled to AFTER ASIC

Laser beams

Dark noise hit

Dark noise hit

~60 PE

~20PE~5 PE

~60 PE

9

Fulfilling the dynamic range and energy resolution requirements

For calibration need to identify 1 PE peak• Noise set to 0.2 PE

Maximum dynamic range = 400 PE• After-pulsing may

increase beyond 400 pixel

ASIC noise ~ 2,000 e-• MPPC gain ~ 5 105

• 0.2 PE noise ⇒attenuate by ~50

ASIC dynamic range = 600 fC• Dynamic range 0.2 PE

to ~200 PE • Need another channel

with higher attenuation

10

Coupling AFTER ASIC to MPPC

Issues• Attenuation

High/low input to ASIC• Low input capacitance

Low electronic noise• Noise from resistors• MPPC recovery

Require small Rbias with purely capacitive termination

• Minimize reflections (50 Ω line)• Pulse shape

Solution• Not clear yet. Some answer

from Spice simulations• Building a specific 8 channel

prototype

MPPC

Charge pump70 V

10-15 cm

DAC-5V to +5V

Charge sum

Rbias ~ 100 kΩ

Cdec = 1 nF

AFTERASIC

Clow = 1 pF

Chigh = 10 pF

Cgnd = 40 pF

Rgnd = 10 kΩ

Rhigh = 0 Ω

Rlow = 0 Ω

Values of R and Care only indicative

D. Bishop, L. Kurchaninov, K. Mizouchi (TRIUMF)

11

Pulse shape and recovery

MPPC

70 V

10-15 cm

Rbias 100 kΩ

Cdec = 1 nF

Chigh = 3.3 pF

Cgnd = 100 pF

Rgnd = 10 kΩ

AFTERASIC

100 pF to ground

N. Jain (Darmouth), and T. Lindner (UBC)

12

Timing resolutionObtained by fitting waveforms• Fit rising edge only

Source of fluctuations• Photon arrival time

Fiber and scintillator decay constants

Waveform distortion• Dark noise• After-pulsing

Need to measure after-pulsing to evaluate effect

4±1 nsData + rising edge fit

5 nsData + full waveform fit

3 nsSimulations + waveform fit

Resolution for MIP (20 PE)

Configuration

T. Lindner, S. Oser (UBC)

13

Beyond the gross featuresEstimating the MPPC NuisancesDark noise• Add pulses. Increase data size

But useful for gain calibration• At <500 kHz, does not affect timing and energy

resolutionCross-talk• Marginal worsening of energy resolution (if <20%)• Increase number of PE

May skew timing resolution

After-pulsing• Worsen timing resolution when fitting full waveform

14

Measuring Dark noise, cross-talk and after-pulsing

Fast recovery biasing scheme: no resistance in seriesTrigger on Dark noise hits (~0.3 PE threshold)Use fast amplifier (CAEN N978)Use 1 GHz digitizer (CAEN V1789)Search for pulses• Extract MPPC*Amplifier response function• Search for pulses based on rise time + fall time +

amplitude criteria• Fit by a superposition of response functions

Add more pulses if poor fit (partial pulse overlap)• Pulse finding is the main source of systematic errors

15

Typical waveforms with after-pulsing test setup

2 missedpulses

Pulse addeappropriate

2.0 PE (cross-talk)

∏ Data (70V, 25C) pulse finder∏ First pass refit∏ Refit after splitting

1.06 PE @ 6490.96 PE @ 6530.90 PE @ 6693.21 PE @ 719

16

Amplitude vs timefor all pulses

Trigger pulseCross-talk = 1-N(1PE)/Ntot

70 V, 25C

17

Hit amplitude vs time70 V, 25C

Expected 8.75 ns recoverytime constant ⇒ could be lengthened

Trigger pulses

Cross-talk region

18

Reducing after-pulsing by playing with recovery time

It is possible to reduce after-pulsing by increasing the recovery time• Resistance in series with bias• Introduce dead time after the pulse

Is there an acceptable compromise?

• For the FGD, readout issue may force us to run with a long recovery time

After-pulsing is then automatically reduced

FGD approach• Run a low bias voltage: after-pulsing ~ 10%

19

Separating Dark Noise and after-pulsing

Count all hits• No cross-talk• Sensitive to

multiple after-pulse

Histogram the time of the 1st

hit after trigger• No cross-talk• No multiples

But more complicated fit

70VAll hits

69.5VAll hits

70 V1st hit after trigger

69.5 V1st hit after trigger

2 exponential fit1 exponential fit

20

After-pulsing fit results

Fit is impaired by low statistics• 69.5V and 70V have

more statistics • Long time constant

hard to pin downIncrease of constant in all hits expected

• Short time constant 20-30 ns

Dominate the after-pulsing

21

Competing contributions

Total after-pulsing

Is dark noise really saturating andthe visible increase due to after-pulsing?

22

Conclusions

MPPC + AFTER combination fulfill FGD requirementsMPPC nuisances are under control for the FGD application• After-pulsing is dominant

Run MPPC at low bias to avoid significant after-pulsingNot a problem. Quantum efficiency is large enough

Investigating interplay between recovery, pulse shape, and after-pulsing• Is there an optimum design?

23

Back-up

24

Measuring after-pulsing with gate technique

B. Kirby (UBC)

25

Measuring after-pulsing with average

techniqueAverage of all pulses1PE response function

Dark noise contribution (fit)

Average after dark noise subtractionAfter-pulsing contribution (fit)Cross-talk contribution (fit)

R. Tacik (U. Regina)

26

1st hit timing distribution fit function

tDNtt

tDN eeApDNeApApedtdN ⋅−−−

⋅− +⎥⎦

⎤⎢⎣

⎡⋅−−= ττ

τ)1(*/

DN = dark noise rateAp = After-pulsing probabilityτ = After-pulsing time constant