Study of Future 3D Calorimetry Based on LYSO or LaBrCe

FutureLYSO/LaBrCeCalorimetry

P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Study of Future 3D Calorimetry Based onLYSO or LaBrCe Crystals for High Precision

Physics

Patrick [email protected]

Paul Scherrer Institut

Zurich PhD Seminar9.10.2019

P. Schwendimann (PSI) Future LYSO/LaBrCe Calorimetry Zurich PhD Seminar, 9.10.19 1 / 20


P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Overview

1 A Brief Introduction

2 A Word on the Simulation

3 Simulation Results

4 Conclusions Towards a Prototype



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

The Quest of High Precision Particle Physics

Find evidence for BSM physics in testing SM predictions to anunprecedented accuracy.

Charged Lepton Flavour Violation [1, 2]

Looking for decays like µ→ eγ, µ→ eee etc.

Prohibited by the SM, highly suppressed by light neutrinos.

Predicted by some BSM theories.

The Legendary Needle in the Haystack: µ→ eγ [3, 4]

Signal

µe

γ

BR(µ→ eγ) > 6 · 10−14 ???

Radiative Muon Decay

µeνe

νµγ

Accidental Background

µeνe

νµγ

Racc ∝ R2µ ·∆E2

γ ·∆Pe ·∆Θ2eγ ×∆teγ



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Measuring Energy Deposits

Calorimetry

Measure the energy a particle deposits in the detector. If it islarge enough, the total energy of the particle will be deposited.

Scintillator

Scintillators emit visible light when hit by high energy radiation.The number of photons is roughly proportional to the energydeposit inside the scintillator.

Silicon Photo-Multipliers (SiPMs)

Semiconducting device to detect optical photons. They consistof thousands of pixels - each of them can be fired by a singlephoton.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

The Idea in General

Scintillator

Quartz

SiPMs

PCB

Carbon Fibres

Goal: Detect Photons of O(50 MeV)

Build a calorimeter by attaching SiPMs to a scintillator.

Thin SiPMs allow readout on front and back.

Use granularity for geometrical reconstruction.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

What kind of Scintillator to use?

Selection of Scintillating Materials

Densityρ (g/cm3)

Light YieldLY

(ph/keV)

Decay Timeτ (ns)

RadiationLenthX0 (cm)

LaBr3(Ce) 5.08 63 16 2.1LYSO 7.1 27 41 1.21NaI(Tl) 3.67 38 245 2.59BGO 7.13 9 300 1.12

Properties

Light Yield: Amount of visible light emitted per energy depositDecay Time: Time scale on which the light gets emittedRadiation Length: Depth scale on which the energy is deposited



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

The MC Simulation

GEANT4 based using custom code for:

SiPM

Include approximate geometry. Create active area physicalvolume, not single pixels

Determine pixel hit based on mathematical formula.

Include quantum efficiency check.

Waveform Generation

Use real SiPM and DAQ response to 1 photoelectron.

Sum detected photons passing dead time check, add noise.

Extract charge and time for each channel.

Time extraction on each side on sum of waveforms.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Crystal Sizes Investigated

”Available”

These crystal sizes were promised by industry:LaBr3(Ce) 20 cm length, 9 cm diameterLYSO 16 cm length, 7 cm diameter

”Large”

Crystals with increased diameter:LaBr3(Ce) 20 cm length, 15 cm diameterLYSO 16 cm length, 15 cm diameter

”Ultimate”

Crystals with 10 Moliere radii and 15 radiation lengths:LaBr3(Ce) 31.5 cm length, 46 cm diameterLYSO 17 cm length, 40 cm diameter



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Variable Estimation Algorithms

Charge: Sum integrated charge over all channels

Qtot =∑

qi

Time: Estimate from front and back times tf , tb (next slide)

t =(n− 1)tf + (n+ 1)tb − L/c(n2 + n)

2n

Position: Estimate from times ti, charges Qi and theiraveraged position xi

x = axxf + bxxb + cx

z = az ln(Qf ) + bz ln(Qb) + cztf + dztb + ez

Parameters trained on the first 10 % of each configuration.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Time Extraction Algorithms

3 3.5 4 4.5 5 5.5 6 6.5 7 (in ns)

0t

0

200

400

600

800

1000

1200

1400

1600

coun

ts Time Reconstruction (All) = 0.0310(3) nsσ

Time Reconstruction (Square) = 0.0428(3) nsσ

Time Reconstruction (Intense) = 0.0419(3) nsσ

Time Reconstruction (First) = 0.0395(3) nsσ

Time Reconstruction (Sum) = 0.0260(2) nsσ

Low Noise

LYSO crystal(R = 3.5 cm, L = 16 cm)

All: average over allchannelsSquare: average over thechannels in a squarearound the one with themost chargeIntense: average over the10 channels with the mostcharge.First: average over the 10first channelsSum: Sum up allwaveforms on each side,then take CF



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Time Extraction Algorithms

3 3.5 4 4.5 5 5.5 6 6.5 7 (in ns)

0t

0

100

200

300

400

500

600

700

800

coun

ts Time Reconstruction (All) = 0.339(4) nsσ

Time Reconstruction (Square) = 0.0736(6) nsσ

Time Reconstruction (Intense) = 0.0716(6) nsσ

Time Reconstruction (First) = 0.0641(5) nsσ

Time Reconstruction (Sum) = 0.0517(4) nsσ

Medium Noise

LYSO crystal(R = 3.5 cm, L = 16 cm)

All: average over allchannelsSquare: average over thechannels in a squarearound the one with themost chargeIntense: average over the10 channels with the mostcharge.First: average over the 10first channelsSum: Sum up allwaveforms on each side,then take CF



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Noise Effects

800 900 1000 1100 1200 1300 1400 1500 1600

310×

q(a.u.)

0

20

40

60

80

100

120

140

coun

tsLow Noise (0.5 pe)

= 0.0167(5) µ/σMed Noise (5 pe)

= 0.0168(5) µ/σHigh Noise (10 pe)

= 0.0171(4) µ/σ

Charge

3 3.2 3.4 3.6 3.8 4 4.2 4.4 4.6 4.8 5t_0 (ns)

0

100

200

300

400

500

600

700

800

coun

ts

Low Noise (0.5 pe) = 0.0260(2) nsσ

Med Noise (5 pe) = 0.0517(4) nsσ

High Noise (10 pe) = 0.0855(6) nsσ

Time

40− 30− 20− 10− 0 10 20 30 40 x (mm)∆

0

50

100

150

200

250

300

coun

ts Low Noise (0.5 pe) = 2.38(5) mmσ

Med Noise (5 pe) = 2.37(6) mmσ

High Noise (10 pe) = 2.46(5) mmσ

x Difference

40− 30− 20− 10− 0 10 20 30 40 z (mm)∆

0

20

40

60

80

100

120

140

160

180

coun

ts Low Noise (0.5 pe) = 4.42(8) mmσ

Med Noise (5 pe) = 5.29(14) mmσ

High Noise (10 pe) = 5.7(2) mmσ

z Difference

min

oreff

ect

min

oreff

ect

sign

ifica

nt

effec

tob

serv

able

effec

t

LYSO crystal (R = 3.5 cm, L = 16 cm)



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Simulations of Available Sizes

800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800310×

q (a.u)

0

50

100

150

200

250

300

350

400

450

Cou

nts

(a.u

.)

Charge

D = 7 cm, L = 16 cm, LYSO = 1.69(6) % µ/σ

:Ce3

D = 9 cm, L = 20 cm, LaBr = 2.52(8) % µ/σ

40− 30− 20− 10− 0 10 20 30 40 z (mm)∆

0

100

200

300

400

500

600

700

Cou

nts

(a.u

.)

z Resolution

D = 7 cm, L = 16 cm, LYSO = 4.39(8) mm σ

:Ce3

D = 9 cm, L = 20 cm, LaBr = 4.17(8) mm σ

Significant differences only for charge resolution. Explainedby larger energy leakage through lateral side in LaBr3(Ce).

Time resolution around 30 ps

Position resolution around 3 mm perpendicular to thecrystal axis



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Spread Out Photons

Consider photons spread out over the whole front face:

800 900 1000 1100 1200 1300 1400 1500

310×

q (a.u.)

0

20

40

60

80

100

120

140

160

180

200

Cou

nts

(a.u

.)Uncut

= 2.67(12) %µ/σMC Truth Cut

= 0.402ε = 2.06(10) % µ/σRec. Pos Cut

= 0.400ε = 2.13(11) % µ/σSkewness Cut

= 0.438ε = 2.06(10) % µ/σUnspread (scaled)

= 1.69(6) %µ/σ

Charge

Reduced charge/energy resolution due to lateral events.Recover to some extend by applying appropriate cuts.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Geometrical Cuts

MC Truth CutUse true position of first interaction as recorded by MCsimulation. Select events with some distance from theborder.

Reconstructed CutUse the reconstructed (x, y)-Position to reject events closeto the border.

Skewness CutAnalyse the radial distribution of the collected charge.Reject events below a certain skewness - most chargecollected on the outer region with distinct tail towards thesmaller radii.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Increasing the Size

800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800310×

q (a.u)

0

200

400

600

800

1000

1200

1400

1600

1800

Cou

nts

(a.u

.)

Charge

D = 7 cm, L = 16 cm, LYSO = 1.69(6) % µ/σ(Ce)

3D = 9 cm, L = 20 cm, LaBr = 2.52(8) % µ/σD = 15 cm, L = 16 cm, LYSO = 0.444(10) % µ/σ

(Ce)3

D = 15 cm, L = 20 cm, LaBr = 0.94(3) % µ/σ

Better energy resolution for larger diameters.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Ultimate Crystals

1000 2000 3000 4000 5000 6000310×

q (a.u)

0

200

400

600

800

1000

1200

Cou

nts(

a.u.

)

Charge

LYSO (R = 20 cm, L = 17 cm) + sensL = 0.242(4) %µ/σ

LYSO (R = 20 cm, L = 17 cm) + Hamamatsu = 0.334(5) %µ/σ

LaBrCe (R = 23 cm, L = 31.5 cm) + sensL = 0.196(4) %µ/σ

LaBrCe (R = 23 cm, L = 31.5 cm) + Hamamatsu = 0.304(5) %µ/σ

Time resolution O(30 ps), Position resolution O(5 mm).



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Industry Issues

”Available” LYSO crystal has defects ...

800 900 1000 1100 1200 1300 1400 1500 1600310×

q (a.u)

0

20

40

60

80

100

120

140

160

180

200

220

240

Cou

nts

(a.u

.)

Charge

D = 7 cm, L = 16 cm = 1.68(5) % µ/σ

D = 7.5 cm, L = 10 cm = 1.78(7) % µ/σ

Closest size: 7.5 cm diam., 10 cm length.No significant decrease in performance.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Summary

Resolutions

LaBr3(Ce) LYSOEnergy (%) 2.5 / 0.9 / 0.3 1.7 / 0.4 / 0.3Time (ps) 28 / 30 /39 26 / 28 /36x Position (mm) 3 / 3.7 /5.7 2.4 / 3.0 / 3.6z Position (mm) 4 / 4.8 /5.4 4.4 / 5 / 6

Values refer to available/large/ultimate crystals.

Conclusion

Light yield is not the limiting factor for the resolutions.

LYSO performs better due to higher density.

Spread out irradiation worsens resolution, geometrical cutsallow some recovery.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

Next Steps

Simulation

Mostly done.

Further runs using the detailed information about theprototype.

Refine analysis algorithms in the final configuration.

The Prototype

Selection based on simulations:

LYSO crystal with 10 cm length and 7.5 cm diameter.

Hamamatsu MPPC S13360-6025PE

Confirming final orders with industry. Assembly to follow soon.



P. Schwendi-mann

Introduction

Simulation

Results

Prospects

References and Further Reading

References

[1] A. M. Baldini et al. arXiv:1812.06540v1 [hep-ex]

[2] G. Cavoto et al. Eur. Phys. J. C 78 (2018), 37

[3] A. M. Baldini et al. (MEG II Collaboration)arXiv:1301.7225v2 [physics.ins-det]

[4] A. M. Baldini et al. (MEG II Collaboration)Eur. Phys. J. C 78 (2018), 380

Further Reading

A. Papa, P. Schwendimann, NIM A 936 (2018), 130


Documents

Study of Future 3D Calorimetry Based on LYSO or LaBrCe