SiW ECAL R&D in CALICE Nigel Watson Birmingham University For the CALICE Collab. Motivation CALICE Testbeam Calibration Response/Resolution MAPS Option Summary

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Motivation for Si/W Shower containment in ECAL, X 0 large Small R moliere and X 0 – compact and narrow showers int /X 0 large, EM showers early, hadronic showers late ECAL, HCAL inside coil Lateral separation of neutral/charged particles/particle flow Strong B field suppresses large beam-related background in detector Compact ECAL (cost of coil) Tungsten passive absorber ~1cm 2 silicon pixel readout, minimal interlayer gaps, stability Studying ~50 m MAPS pixels as swap-in option, e.g in SiD, ILD, CLICnn? CMOS process, more mainstream Industry standard, multiple vendors (schedule, cost) (At least) as performant – ongoing studies Simpler assembly Power consumption larger – but better thermal properties

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct CALICE: From MC to Reality to MC Initial task Build prototype calorimeters to establish viable technologies and compare objectively Collect hadronic shower data with unprecedented granularity tune reconstruction algorithms validate existing MC models Ultimate goal High granularity calorimeter optimised for Particle Flow measurement of multi-jet final state at the ILC (or CLIC or …) CAlorimeter for the LInear Collider Experiment Imaging calorimeter Next task Exploit validated models for whole detector optimisation Next task Exploit validated models for whole detector optimisation

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct CALICE Test Beam Prototypes 1x1cm 2 lateral segmentation 1 X 0 longitudinal segment. ~1 total material, ~24 X 0 3x3cm 2 tiles lateral segmentation ~4.5 in 38 layers 5x100cm 2 strips ~5 in 16 layer 10 GeV pion CERN test beam 10 GeV pion CERN test beam SiW ECAL Scint-Fe HCAL Scint-Fe tail catcher/ muon tracker Scint-Fe tail catcher/ muon tracker beam See talk by Felix Sefkow

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct The 2006 CERN Testbeam HCAL Tail Catcher ECAL beam SiW ECAL 30x30x20cm 6.4k channels (9.8k in 2008) SiW ECAL 30x30x20cm 6.4k channels (9.8k in 2008) AHCAL layer with high granular core readout

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct ECAL Prototype Overview 62 mm 20cm 36cm 30 layers, 3 tungsten thicknesses Active silicon layers interleaved Very Front End chip / readout on PCB W layers wrapped in carbon fibre PCB+Si layers:8.5 mm 6x6 1x1cm 2 Si pads Conductively glued to PCB 14 layer PCB, VFE analogue signals DAQ

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Mechanical Structure for TestBeam Differing W absorber thicknessOptional indexed offsets between stacks

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Production & Testing Mounting/gluing the wafers Using a frame of tungsten wires 6 active silicon wafers 12 VFE chips 2 calibration switch chips Line Buffers To DAQ PCB designed in LAL-Orsay, made in Korea (KNU) 60 Required for Prototype Automation, glue : EPO-TEK® EE129-4 Glue/place ( 0.1 mm) of 270 wafers with 6×6 pads 9720 glue dots Production line set up at LLR

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Real Detector Effects Significant part of R&D is understanding which effects are important to the measurement What details should be simulated, e.g. Non-uniformity of passive material Essential to include this level of realism in models What can be corrected a posteriori Inter-wafer gaps (guard rings) To obtain uniform response What has to be redesigned Guard ring scheme (square events)

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Pedestals and Noise Performance Residual pedestals in non-beam events after all known effects accounted for Gaussian fits, channel-by-channel Uniformity in pedestals Residual offset=0.2% MIP Channel-channel = 1.7±0.1% MIP Run-run = 1.1±0.4% Noise / channel 12.9±0.1% MIP 94% channels with run-run variation<5% Noise Pedestals

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct CalibrationCalibration 74 x 250k beam halo muon runs Gaussian x Landau fits, channel- by-channel to extract calibration constant (most prob.value) Uniformity across channels 30 GeV

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Cross-Talk and Mitigation Large quantity energy deposited close to guard rings causes ~constant amplitude signal in distinct square pattern Consequence of capacitative coupling between guard rings and peripheral diode pads Simulation model supports hypothesis By segmenting guard rings, expect reduction in effect by factor x3-30 Example of improved design only by building prototype

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Electron Event Selection Simple cuts Based on raw energy sum Cerenkov rejects intermediate Rejection of pre- shower events Beam halo on run- by-run basis E raw Preliminary +Cerenkov

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Event Selection Data sample from CERN 2006 testbeam used in results below Much larger samples from 2007/8 runs at CERN/FNAL Also ±, ±, p data Future will include combined analysis of data from individual CALICE detector subsystems Preliminary

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Inter-Wafer Gaps 15% 20% Response loss wider and less deep in x x layers staggered y layers aligned Gaps dominated by 1mm guard rings around each 6x6 wafer Preliminary

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Inter-Wafer Gaps Statistical correction for unmeasured energy in gaps (~7% area) Response function Smooths response Some cost in resolution Low energy tail in observed energy much improved 20 GeV e - Preliminary

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Sampling Fraction Precise detector, sensitive to even small effects Odd-even layer asymmetry at 7% level Arises due to small differences in passive material in addition to W absorber (PCB+glue+ Cfi+…) 7% Beam, normal incidence Preliminary

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct ECAL Hit Energy, 30 GeV e - Hit energy ECAL energy/hit E hit /MIPs Mean/resolution from gaussian fit, each energy Total ECAL energy/event Do include odd/even effects Do not correct for gaps Avoid by fiducial selection Some effects not 100% understood Low energy excess (below MIP peak) Only minor effect on total energy Preliminary

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct ECAL Longitudinal Profile Shower max layer # Beam energy / GeV Cos(angle) Layer # (by W depth)) Energy/layer/event (GeV) Solid: G4/Mokka Dashed: data Expected logarithmic behaviour of shower max.and angular dependence Small deviations data/MC attributed to preshowering upstream of ECAL

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Energy Response, Linearity Energy resolution E meas vs. E beam Non-linearities ~1% Preliminary

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct MAPS ECAL: Option Summary How small? EM shower core density at 500GeV is ~100/mm 2 Pixels must be< m 2 Our baseline is m 2 Gives ~10 12 pixels for ECAL – Tera-pixel APS Mandatory to integrate electronics on sensor How small? EM shower core density at 500GeV is ~100/mm 2 Pixels must be< m 2 Our baseline is m 2 Gives ~10 12 pixels for ECAL – Tera-pixel APS Mandatory to integrate electronics on sensor Swap ~0.5x0.5 cm 2 Si pads with small pixels Small := at most one particle/pixel 1-bit ADC/pixel, i.e. Digital ECAL Effect of pixel size 50 m 100 m >1 particle/ pixel Incoming photon energy (GeV) Weighted no. pixels/event 12 m

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct TPAC1.0 Overview 8.2 million transistors pixels; 50 m; 4 variants Pixel: 4 diodes, Q-preamp, mask+trim Sensitive area 79.4mm 2 Four columns of logic+SRAM Logic columns serve 42 pixel region Hit locations & (13 bit) timestamps Local SRAM 11% deadspace for readout/logic Data readout Slow (<5 MHz) – train buffer Current sense amplifiers Column multiplex 30 bit parallel data output Region Group (region=7 groups of 6 pixels) Logic/SRAM columns

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct TPAC1.0 Overview 8.2 million transistors pixels; 50 m; 4 variants Pixel: 4 diodes, Q-preamp, mask+trim Sensitive area 79.4mm 2 Four columns of logic+SRAM Logic columns serve 42 pixel region Hit locations & (13 bit) timestamps Local SRAM 11% deadspace for readout/logic Data readout Slow (<5 MHz) – train buffer Current sense amplifiers Column multiplex 30 bit parallel data output Region Group (region=7 groups of 6 pixels) Logic/SRAM columns

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Beam Background Beam-beam interaction by GUINEAPIG LDC01sc (Mokka) 2 machine scenarios : 500 GeV baseline, 1 TeV high luminosity purple = innermost endcap radius 500 ns reset time ~ 2 inactive pixels [O.Miller] To repeat in SiD01, CLICnn, verify optimisation To repeat in SiD01, CLICnn, verify optimisation X (mm) y (mm) 1TeV high lumi ECAL endcap hits

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Single Pixel Characterisation: Laser Stimulus F B Pixel profile Charge collecting diodes 50 m Amplitude results With/without deep p-well Compare Simulations - GDS Measurements - Real

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Single Pixel Characterisation: 55 Fe source 55 Fe gives 5.9keV photon Deposits all energy in point in silicon: 1640 e Sometimes maximum energy deposited in a single diode without diffusion absolute calibration! Binary readout from pixel array Derivative of distribution to get signal peak in threshold units (TU)

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct MAPS Outlook MAPS ECAL: alternative to baseline analogue SiW Multi-vendors, potential cost/performance gains New INMAPS deep p-well process (optimise charge collection) Basic physics benchmark studies (no harm) to evaluate performance relative to baseline designs for future LC detectors First Sensor, TPAC 1.0 Four sensor architecture variants on 9x9mm 2 device Successful operation of highly complex pixels See 55 Fe, laser charge injection, beam particles Proved viability of the Deep P-Well Revised Sensor, TPAC 1.1 – received from fab. Oct Homogeneous 28k pixel array Pin- and form factor compatible with original sensor Full characterisation starting (~ one week) Testbeam with single particles Spring 2009 No show stoppers, continue concept for DECAL Will consider in any detector concept / accelerator Future plans TPAC 2.0, full reticle size (2.5x2.5cm 2 ) sensor Multiple layer detector, contained e.m. showers Proof of principle demonstration of digital ECAL resolution/linearity

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct CALICE is developing exceptionally performant calorimetry for ILC (+CLIC+…) Integrated approach, controlled technology evaluatation (Sefkow, Blaha) Analogue SiW – baseline technology used by SiD, ILD MAPS SiW (not this talk) scintillator ECAL, testbeam at FNAL Sept./Oct./ 08 First CALICE ECAL paper published, 2008_JINST_3_P08001 Detailed investigation of technical performance of physics prototype 9760 channel, 24 X 0 ECAL - calibration, stability, design of DAQ, … Large amounts of data collected at DESY/CERN SPS/FNAL MTEST Papers on transverse/longitudinal profile, technology and hadronic model testing,… in progress Improving on lessons learned, e.g. guard rings Developing next-generation prototypes within the EUDET framework - realistic ECAL and HCAL modules Use experience from modelling test beam prototypes to add appropriate realism to whole detector concept models Reduce uncertainties See CALICE web for further details! SummarySummary

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Backup/sparesBackup/spares

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct SiD 16mm 2 area cells ZOOM μm 2 MAPS pixels Tracking calorimeter

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct CALICE INMAPS TPAC1 Architecture-specific analogue circuitry 4 diodes Ø 1.8 m First round, four architectures/chip (common comparator+readout logic) INMAPS process: deep p-well implant 1 μm thick under electronics n-well, improves charge collection 0.18 m feature size

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct A Tera-Pixel ECAL is challenging Benefits No readout chips CMOS is well-known and readily available Ability to make thin layers Current sources of concern DAQ needs Power consumption/Cooling System considerations

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct DAQ requirements O(10 12 ) channels are a lot... Physics rate is not the limiting factor Beam background and Noise will dominate Assuming 2625 bunches and 32 bits per Hit 10 6 Noise hits per bunch ~O(1000) Hits from Beam background per bunch (estimated from GuineaPIG) Per bunch train ~80 Gigabit / 10 Gigabyte Readout speed required 400 Gigabit/s CDF SVX-II can do 144 Gigabit/s already

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Cooling and power Cooling for the ECAL is a general issue Power Savings due to Duty Cycle (1%) Target Value for existing ECAL ASICS 4 µW/mm2 Current Consumption of MAPS ECAL: 40 µW/mm2 depending on pixel architecture TPAC1 not optimized at all for power consumption Compared to analog pad ECAL Factor 1000 more Channels Factor 10 more power Advantage: Heat load is spread evenly

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct [Marcel Stanitzki] Thermal properties

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Simple analogue vs. digital ECAL resolution Number of charged particles is an intrinsically better measure than the energy deposited Energy deposited (analogue ECAL) resolution ~50% worse than number of particles (digital ECAL) resolution Can we measure the number of charged particles directly? Possible to approach ideal analogue resolution using low noise electronics Is ideal resolution for the digital case achievable?

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Event display Shower from 40 GeV + 20 GeV + HCAL only Clear structure visible in hadronic showerBack-scattered particle

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct CERN Testbeam ECAL: 54 PCBs (30 layers) 216 channels/PCB in central part and 108/PCB in bottom part Total channels: 9072 Total radiation length: 24 X 0 AHCAL: 38 fully commissioned modules 30 modules with fine granularity = 216 tiles 8 modules with coarse granularity = 141 tiles Total channels: 7608 Total interaction length: 4.5 TCMT: 16 layers – fully instrumented Alternated cassettes (from layer 2 to 16) have been staggered in X and Y layer 2 = nominal; layer 3 (vert) = -1 inch in X; layer 4 (hor) = +1 inch in Y; repeated up to layer 16

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct beamline instrumentation description 500 m beam 3x3 10x10 trigger * all in mm News on the beamline 1) Cherenkov operated for e/ and /p separation 2) 3 x/y pairs of MWPC with double readout 3) 10x10 cm trigger only (no 3x3) 4) amplitude r/o of 1cm thick scint. counter (20x20 inner veto) + outer veto with 20x20 cm hole to tag double particle 5) hodoscope installed for initial muon runs and from ECAL chip irradiation to end hodoscope 1 m

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Energy points and particle types Proposed in TB planCollected during TB Energy (GeV) 6,8,10,12,15,18,20,25,30,40,50, 60,80 6,8,10,12,15,18,20,25,30,4 0,50,60,80,100,120,130,15 0,180 Particles ± /e ± ± /e ± /protons Beam energies extrapolated from secondary beam Electron beam obtained sending secondary beam on Pb target /e separation achieved using Cherenkov threshold detector filled with He gas Possible to distinguish from e for energies from 25 to 6 GeV /proton separation achieved using Cherenkov threshold detector with N 2 gas Possible to distinguish from protons for energies from 80 to 30 GeV

Nigel Watson / BirminghamCLIC'08 Workshop, CERN, 15-Oct Total events collected Integrated Luminosity Event Types