Beam-tests of prototype modules for the CMS High Granularity - - PowerPoint PPT Presentation

▶

Aug 20, 2022 179 likes •501 views

Motivation for upgrade HGCAL prototype Preliminary results Conclusion Beam-tests of prototype modules for the CMS High Granularity Calorimeter at CERN PIXEL2018: International Workshop on Semiconductor Pixel Detectors for Particles and

SLIDE 1

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Beam-tests of prototype modules for the CMS High Granularity Calorimeter at CERN

PIXEL2018: International Workshop on Semiconductor Pixel Detectors for Particles and Imaging 2018

Arnaud Steen On behalf of the CMS collaboration 10-14 December 2018

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 1 / 18

SLIDE 2

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Outline

Motivation for upgrade

HGCAL prototype

Preliminary results

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 2 / 18

SLIDE 3

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

CMS will replace its endcap calorimeters in 2024-25

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 3 / 18

SLIDE 4

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

High luminosity LHC

High lumi LHC

Luminosity : 5 × 1034 cm−2s−1 Integrated luminosity : 3000 fb−1 Fluences : up to 1016neq/cm2 in ECAL endcap Average pileup : 140 (maximum of 200)

Need to replace endcap calorimeters

Radiation tolerant Good timing resolution (pileup mitigation) Tracking capability (shower reconstruction, PFA)

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 4 / 18

SLIDE 5

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

CMS High Granular Calorimeter (HGCAL)

More details about HGCAL project in Chia-Hung Chien’s poster : The CMS High Granularity Calorimeter for HL-LHC

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 5 / 18

SLIDE 6

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Overview and goals of the beam tests

Main goals

Validation of basic design of the HGCAL Comparisons with GEANT4 simulation Test calorimetric performance of silicon based calorimeter

2016 beam tests

Timing study with irradiated silicon diodes at CERN Electromagnetic calorimeter prototype using 6" modules and SKIROC2 ASIC (designed for CALICE SiWECal) at FNAL and CERN

2017/2018 beam tests

New 6" modules using SKIROC2_CMS ASIC (with time over threshold and time of arrival measurements) Various configurations tested at CERN with up to 20 modules in 2017 Beam test at DESY (March/April 2018) with few modules Larger scale beam tests :

◮ CE-E prototype with 28 modules at CERN (June 2018) ◮ Prototype with 94 silicon modules + the CALICE Analog Hadronic CALorimeter prototype

(sampling calorimeter using scintillator tiles as active medium) at CERN (October 2018)

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 6 / 18

SLIDE 7

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Results from 2016 beam tests

Precision timing study with irradiated silicon diodes Electromagnetic calorimeter prototype using 6" modules and SKIROC2 chip (designed for CALICE SIWECal)

◮ FNAL beam test : 16 modules with total

thickness ≈ 15X0

◮ CERN beam test : 8 modules, with total

thickness ≈ 27X0

Beam tests results in this paper :

N. Akchurin et al. First beam tests of

prototype silicon modules for the CMS High Granularity Endcap Calorimeter (link)

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 7 / 18

SLIDE 8

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Results from 2016 beam tests

Precision timing study with irradiated silicon diodes Electromagnetic calorimeter prototype using 6" modules and SKIROC2 chip (designed for CALICE SIWECal)

◮ FNAL beam test : 16 modules with total

thickness ≈ 15X0

◮ CERN beam test : 8 modules, with total

thickness ≈ 27X0

Beam tests results in this paper :

N. Akchurin et al. First beam tests of

prototype silicon modules for the CMS High Granularity Endcap Calorimeter (link)

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 7 / 18

SLIDE 9

Motivation for upgrade HGCAL prototype Preliminary results Conclusion 1

Motivation for upgrade

HGCAL prototype

Preliminary results

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 7 / 18

SLIDE 10

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Module assembly

Glued stack

Baseplate (CuW for CE-E, Cu for CE-H), thickness : 1.2 mm Kapton foil with gold layer 6" silicon sensor from HPK PCB holding 4 SKIROC2_CMS readout chips (32 channels per chip are connected to a silicon pads) Silicon pads wire bonded through holes in the PCB

Silicon sensors

6" hexagonal sensor Physical thickness : 320 µm Depleted thickness : 300 µm (4 modules with 200 µm) 134 individual silicon cells with an area

f ≈1.1cm2 for the full cells

2 inner calibration pads with 1/9th of the area of the full cell (for MIP sensitivity after irradiation)

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 8 / 18

SLIDE 11

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Readout CHIP : SKIROC2_CMS

Derived from CALICE SKIROC2 chip 64 channels (though 32 are connected to a silicon pad) 2 slow shapers (shaping time between 10 and 70 ns) per channel with

◮ a 13-deep 40 MHz analog memory used

as waveform sampler

◮ 12-bit ADC

Fast shaper (shaping time between 2 and 5 ns) and discriminators for

◮ Time over threshold to measure large

signal (preamplifier saturation region)

◮ Time of arrival (50 ps timing resolution

foreseen)

More details in this paper : J. Borg et al. SKIROC2_CMS an ASIC for testing CMS HGCAL

Overview of SK2_CMS

64 channels versatile Si calorimeter readout based on CALICE SKIROC2

time [ns]

50 100 150 200 250 300

Low gain ADC counts

200 − 100 − 100 200 300 400

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 9 / 18

SLIDE 12

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Readout CHIP : SKIROC2_CMS

Derived from CALICE SKIROC2 chip 64 channels (though 32 are connected to a silicon pad) 2 slow shapers (shaping time between 10 and 70 ns) per channel with

◮ a 13-deep 40 MHz analog memory used

as waveform sampler

◮ 12-bit ADC

Fast shaper (shaping time between 2 and 5 ns) and discriminators for

◮ Time over threshold to measure large

signal (preamplifier saturation region)

◮ Time of arrival (50 ps timing resolution

foreseen)

More details in this paper : J. Borg et al. SKIROC2_CMS an ASIC for testing CMS HGCAL

Overview of SK2_CMS

64 channels versatile Si calorimeter readout based on CALICE SKIROC2

Injected charge [MIP]

100 200 300 400 500 600 700 800 900 1000

ADC counts

500 1000 1500 2000 2500 3000 3500

High gain Low gain TOT Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 9 / 18

SLIDE 13

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

CE-E prototype

Mini cassette

4 mm thick lead absorber plate 6 mm thick copper cooling plate holding 2 modules on its 2 faces Copper cooling pipe inside the copper plate Aluminium frame and mylar foil

ECAL prototype

14 mini cassetes (i.e. 28 silicon modules) Total depth : ≈24 X0, 1λI Water cooling to keep constant temperature : 28◦C

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 10 / 18

SLIDE 14

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

CE-H prototype

CE-H layer

6 mm thick copper cooling plate holding 7 modules (only on front face) Copper cooling pipe glued to the backside of the copper plate

CE-H (silicon part) prototype

12 layers (9 with 7 modules + 3 with 1 module), 12k electronic channels in total Separated by 4 cm thick steel absorber plates Total depth : ≈3.5λI Water cooling to keep constant temperature : 28◦C

CALICE AHCAL prototype

38 layers using scintillator tiles (tile size : 3×3cm2) ≈ 22k channels Separated by 1.75 cm thick steel absorber plates Total depth : ≈4λI

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 11 / 18

SLIDE 15

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

DAQ systems

Extra detectors in the beam line :

2 scintillators in coincidence for the trigger + 1 veto scintillator between HGCAL and AHCAL prototypes to reject pion in electron run 4 delay wire chambers readout with TDC 2 MCP detectors for timing reference readout with 5 GHz digitizer

Custom DAQ boards

SYNCH board (1 synch board in total)

◮ board creates 40 MHz clock ◮ receives and distributes triggers to RDOUT boards, AHCAL, TDC and digitizer

RDOUT boards connected to up to 8 modules

◮ distribute low and bias voltage to the modules ◮ send commands to the MAX10 FPGA on the modules (trigger, slow control) ◮ read out the data using IPbus protocol

DAQ software

eudaq framework https://eudaq.github.io/ Collects data from each detector types Provides DQM framework

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 12 / 18

SLIDE 16

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Beam test summary

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 13 / 18

SLIDE 17

Motivation for upgrade HGCAL prototype Preliminary results Conclusion 1

Motivation for upgrade

HGCAL prototype

Preliminary results

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 13 / 18

SLIDE 18

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Analysis procedure

Pedestal estimation and subtraction for each channel and each memory slot (13-deep 40 MHz analog memory) Common mode noise estimation for each module and each time sample Pulse shape fit to extract high and low gain amplitudes, after hit pre-selection Gain choice between high gain, low gain and ToT and hit energy calibration

◮ High gain calibrated using muon runs ◮ 1 MIP ≈ 40-50 ADC counts (high gain),

S/N≈6-7 for MIP

◮ Low gain calibrated using high gain ◮ ToT calibrated using low gain

266 262 263 249 272 265 251 258 272 295 265 269 260 261 254 242 265 266 263 264 266 263 258 255 238 242 247 243 263 264 266 269 277 267 240 241 229 251 238 247 264 262 262 249 248 251 239 243 248 241 257 263 262 236 256 258 247 240 248 243 247 240 241 251 241 249 251 252 247 240 256 244 241 241 253 259 244 253 249 247 237 248 253 252 250 269 257 245 242 246 256 237 255 263 251 260 253 247 251 243 249 249 244 257 262 270 250 252 263 263 243 263 248 253 265 254 265 245 268 263 241 268 248 259 269 257 270

X [cm]

6 − 4 − 2 − 2 4 6

Y [cm]

6 − 4 − 2 − 2 4 6

High gain [ADC counts]

200 210 220 230 240 250 260 Pedestal MAP

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 14 / 18

SLIDE 19

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Analysis procedure

Pedestal estimation and subtraction for each channel and each memory slot (13-deep 40 MHz analog memory) Common mode noise estimation for each module and each time sample Pulse shape fit to extract high and low gain amplitudes, after hit pre-selection Gain choice between high gain, low gain and ToT and hit energy calibration

◮ High gain calibrated using muon runs ◮ 1 MIP ≈ 40-50 ADC counts (high gain),

S/N≈6-7 for MIP

◮ Low gain calibrated using high gain ◮ ToT calibrated using low gain

7.5 8.5 10.5 10.5 11.5 8 8 10.5 11 11.5 10.5 11 10.5 8 11 10.5 11.5 10.5 12 12 13.5 9.5 1 6 7 11 12 10.5 11.5 11 11 11.5 9.5 9.5 12 10.5 11 10.5 11 11 11 11 10 8.5 8.5 9 8.5 11 12.5 11.5 11.5 11.5 11 9 10.5 10.5 10.5 7.5 12 11.5 11.5 11.5 10.5 11 10 9.5 10 11 9.5 8 11 11 11.5 10 10.5 9 9 10.5 9.5 9 7 10 11 10 10 11.5 9 10.5 10 9.5 10 9 6.5 10 9.5 12 10.5 10 9.5 8.5 10 10.5 8 10.5 10.5 9.5 11 9.5 8 9.5 8.5 8.5 6.5 7.5 9 9.5 10.5 9.5 10 10 7 7.5 9 10 10.5 7.5 6

X [cm]

6 − 4 − 2 − 2 4 6

Y [cm]

6 − 4 − 2 − 2 4 6

High gain [ADC counts]

2 4 6 8 10 12 14 16 18 20

Noise MAP (before CM subtraction)

4.57476 5.55967 5.93032 5.93032 7.4129 5.93032 5.55967 5.93032 6.67161 5.93032 5.93032 5.93032 5.18903 5.93032 5.93032 5.93032 5.93032 5.93032 5.93032 5.93032 6.67161 5.93032 8.15419 4.41689 4.85785 5.93032 5.93032 5.93032 6.67161 6.67161 5.93032 5.93032 5.93032 7.4129 8.15419 5.93032 5.93032 5.93032 5.93032 5.93032 6.67161 5.93032 5.93032 5.93032 5.93032 5.93032 5.55967 5.93032 5.93032 5.93032 5.93032 5.93032 5.93032 5.18903 5.93032 5.93032 5.18903 4.81838 5.93032 5.93032 6.67161 5.18903 5.93032 5.93032 5.93032 5.93032 5.18903 5.93032 5.93032 4.44774 8.15419 5.93032 5.93032 6.67161 5.93032 5.93032 5.93032 5.18903 5.18903 5.93032 4.44774 5.93032 5.93032 10.3781 5.93032 5.93032 5.93032 5.18903 5.18903 5.93032 5.93032 5.93032 4.7154 5.93032 5.93032 5.93032 5.93032 5.93032 5.93032 5.93032 5.18903 5.18903 5.93032 5.93032 5.93032 5.93032 5.18903 5.18903 5.93032 5.93032 5.93032 5.93032 4.18007 5.18903 5.93032 5.93032 5.93032 5.93032 6.67161 6.67161 5.18903 5.18903 6.67161 5.93032 6.67161 5.18903 4.77891

X [cm]

6 − 4 − 2 − 2 4 6

Y [cm]

6 − 4 − 2 − 2 4 6

High gain [ADC counts]

2 4 6 8 10 12 14 16 18 20

Noise MAP (after CM subtraction) Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 14 / 18

SLIDE 20

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Analysis procedure

Pedestal estimation and subtraction for each channel and each memory slot (13-deep 40 MHz analog memory) Common mode noise estimation for each module and each time sample Pulse shape fit to extract high and low gain amplitudes, after hit pre-selection Gain choice between high gain, low gain and ToT and hit energy calibration

◮ High gain calibrated using muon runs ◮ 1 MIP ≈ 40-50 ADC counts (high gain),

S/N≈6-7 for MIP

◮ Low gain calibrated using high gain ◮ ToT calibrated using low gain time [ns]

50 100 150 200 250 300

High gain ADC counts

600 − 400 − 200 − 200 400 600 0.16278 ± = 81.979

max

T 4.2859 ± = 606.84

max

time [ns]

50 100 150 200 250 300

Low gain ADC counts

60 − 40 − 20 − 20 40 60 80 0.49527 ± = 77.602

max

T 1.5661 ± = 73.787

max

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 14 / 18

SLIDE 21

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Analysis procedure

Pedestal estimation and subtraction for each channel and each memory slot (13-deep 40 MHz analog memory) Common mode noise estimation for each module and each time sample Pulse shape fit to extract high and low gain amplitudes, after hit pre-selection Gain choice between high gain, low gain and ToT and hit energy calibration

◮ High gain calibrated using muon runs ◮ 1 MIP ≈ 40-50 ADC counts (high gain),

S/N≈6-7 for MIP

◮ Low gain calibrated using high gain ◮ ToT calibrated using low gain Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 14 / 18

SLIDE 22

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Analysis procedure

Pedestal estimation and subtraction for each channel and each memory slot (13-deep 40 MHz analog memory) Common mode noise estimation for each module and each time sample Pulse shape fit to extract high and low gain amplitudes, after hit pre-selection Gain choice between high gain, low gain and ToT and hit energy calibration

◮ High gain calibrated using muon runs ◮ 1 MIP ≈ 40-50 ADC counts (high gain),

S/N≈6-7 for MIP

◮ Low gain calibrated using high gain ◮ ToT calibrated using low gain Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 14 / 18

SLIDE 23

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Event displays : 300 GeV EM showers

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 15 / 18

SLIDE 24

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Event displays : 300 GeV hadronic showers

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 16 / 18

SLIDE 25

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Electromagnetic shower results

Energy sum from all layers (only CE-E prototype i.e. 28 modules) Beam energy corrected to take into account synchrotron radiation Simulation being tuned with momentum spread from SPS

Electron momentum [GeV/c]

100 200 300

[MIPs] E

10000 20000 30000

Simulation TB Data, October 2018

Electron momentum [GeV/c]

100 200 300

[%] E / σ

2 4 6

Simulation TB Data, October 2018

Next steps :

◮ Improve hit calibration ◮ Apply weights for each layers Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 17 / 18

SLIDE 26

Motivation for upgrade HGCAL prototype Preliminary results Conclusion

Conclusion

New HGCAL prototype has been built with 94 silicon modules (≈ 12k channels) between 2017 and 2018 Combined data taking with AHCAL prototype, delay wire chambers and MCP has been a success Good preliminary results

◮ Low noise ◮ Good S/N for MIP ◮ Reasonable data-simulation agreement ◮ Energy and time resolution close to expectations

Next plan for the data analysis :

◮ Improve the calibration ◮ Compare data and simulation ◮ Apply clustering technique ◮ Combine HGCAL data with AHCAL for hadronic shower study ◮ Time precision study using MCP detectors Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 18 / 18

SLIDE 27

Back-Up

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 18 / 18

SLIDE 28

Calibration of ToA

ToA signal

Using electron runs from June data ToA threshold : ≈10 MIPs ToA measure time between ToA trigger and next falling/rising edge (but skiping first edge) of the 40 MHz clock

ToA calibration

Beam is asynchronous → Time distribution expected to be uniform ToA range = 25 ns (with 12.5 ns offset)

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 18 / 18

SLIDE 29

Time walk correction

Time walk can be large for low signal amplitude Time walk correction :

◮ Select a reference channel (and small amplitude window to minimize TW effect) ◮ Look at time diffence with neighbouring cells to extract the time-walk Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 18 / 18

SLIDE 30

Time resolution

Time differences between 2 silicon pads

◮ Assume same time resolution in each cell → σt = σ∆t /

√ 2

◮ Time resolution constant term ≈ 50 ps

Next step : use MCP data for the time reference

Arnaud Steen, NTU CMS-HGCAL beam-tests 10-14 December 2018 18 / 18