The CMS electromagnetic calorimeter M. Paganoni University of Milano Bicocca and INFN on behalf of the CMS ECAL collaboration The Xth Vienna Conference on Instrumentation Vienna, February 16-21, 2004
M.Paganoni, Univ. Milano Bicocca and INFN The LHC collider pp collider at Ecms=14 TeV max. Luminosity = 1034 cm-2 s-2 inel= 100 mb 109 events/s Higgs= 1 pb 10-2 events/s 1 crossing/25 ns (40 MHz) 20 events/crossing 1000 tracks Doses in ECAL (10 years) Neutrons: 1013 - 1014 n/cm2 Gammas: 4 – 200 kGy Extreme conditions for detectors Granularity (105 107 channels) Speed of response DAQ + trigger (109 102 ev/s) High radiation resistance Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN The CMS experiment Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Higgs hunting at LHC Natural width (GeV) Higgs Mass (GeV) 50 100 200 400 800 0.001 0.004 1.4 30 250 LEP LHC H gg H ZZ* 4l H ZZ 4l H WW or ZZjj With 100 fb-1 and CMS electromagnetic calorimeter Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Physics requirements The benchmark is the discovery of a low mass Higgs: m = 2 E1 E2 (1-cos) H/mH ≤ 10-3 at mH ~100 GeV experimental resolution is crucial Energy resolution : a ~ 0.027 GeV1/2 b < 200 MeV c ~ 0.005 Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
The choice of a crystal calorimeter Excellent energy resolution Structural compactness, easy assembly Tower structure, fine transverse granularity 4.5 2.4 NaI(Tl) BaF2 CsI(Tl) CsI CeF3 BGO PWO r 3.67 4.88 4.53 6.16 7.13 8.26 g/cm3 X0 2.59 2.05 1.85 1.68 1.12 0.89 cm RM 3.4 3.8 2.6 2.2 t 250 0.8/620 1000 20 30 300 15 ns lp 410 220/310 565 310 310/340 480 420 nm n(lp) 1.56 1.80 2.15 2.29 LY 100% 15% 85% 7% 5% 10% 0.2% %Nal 40000 /MeV Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN The choice of PbWO4 Fast scintillation Small X0 and Rm Intrinsic radiation hardness Relatively easy to grow Massive production capability Low Light Yield Strong L.Y. dependence on T Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN CMS ECAL Structure Preshower (Pb/Si) 3 X0 Barrel: |h|<1.48 , 25.8 X0 61200 PbWO4 crystals, ~22x23x230 mm3 Endcap: 1.48 < |h| < 3.0 , 25 X0 14648 PbWO4 crystals, 30x30x220 mm3 Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN ECAL - barrel Module (40/50 Submodules) Submodule (10 crystals) Submodule Module Supermodule (4 modules) Supermodule Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN ECAL - endcaps Supercrystal Dee 138 Supercrystals Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Crystal R&D phase (’95-’98) Satisfactory and homogeneous properties 98 95 Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Crystal Transparency Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Crystal radiation hardness Front irradiation. 1.5 Gy. 0,.15 Gy/h Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Uniformity of light collection all polished Ra=0.34 m Ra=0.24 m Tapered shape of the crystals and high n (2.3) produce non uniformities in the light collection Uniformity can be improved, at the expense of Light Yield, by depolishing one surface with a given roughness Max d(LY)/dX0= 0.35%/X0 LY (8X0) = 8 p.e./ MeV (18 C) Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Crystal production (’01-’04) Technology steps in Bogoroditsk Barrel 2000 65 mm Endcap 1999 44 mm Barrel 1996 32 mm F = 32 mm 2 in one! F = 65 mm Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Check of the crystal quality INFN/ENEA Rome CERN Control and storage to a dbase of: Dimensions Transmission Light Yield and uniformity Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Calorimeter assembly Regional centers in Rome and at CERN 48 modules produced and 8 supermodules assembled Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
The photodetectors (barrel) Avalanche PhotoDiodes developped with Hamamatsu: Solid state detectors: insensitive to the CMS Magnetic Field Good matching to PWO4 emission spectrum (Q.E~80%) and Internal Gain (M=50 for V~380V) compensate for poor L.Y. deff ~ 5 mm (acceptable response to ionizing radiation) dM/dT ~ -2.4% /0C dT/T ~ 0.05 K 1/M dM/dV ~ 3%/V dV/V ~ 10-5 All APD screened for radiation hardness 2 APDs per crystal: 50 mm2 active area Si3N4, SiO2, contact p++ photon conversion p e- acceleration n e- multiplication n- e- drift n++ e- collection contact g Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
The photodetectors (forward) Vacuum PhotoTriodes (fine mesh) in the EndCap Region Rather insensitive to Magnetic Field because of the angle More radiation hard w.r.t APD’s (UV glass window) Gain 8-10 at 4 T ; Q.E.~ 20% at 420 nm ; area ~ 280 mm2 Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Test of a prototype Testbeam on 1999 prototype matrix with 30 preproduction crystals and APD: No sign of rear leakage, with 280 GeV electrons. Fit Resolution as a function of energy: Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Energy resolution: a photostatistics contribution, including : Light Yield light collection efficiency geometrical efficiency of the photodetector photocatode quantum efficiency Npe/GeV = 4000 for 0.5 cm2 APD 1.6% electron current multiplication in APD, contributing a square root of excess noise factor, F = 2 1.61.4 = 2.25% Lateral containment (55 matrix) 1.5% Total stochastic term a = 2.7 % Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Energy resolution: b 40 ns shaping time, summed over 3x3 or 5x5 channels Serial noise (p.d. capacitance) 1/t 150 MeV Parallel noise (p.d. dark current) t, mostly radiation induced negligible at the start of the experiment 30 MeV after 1 year at low luminosity 100 MeV after 1 year at high luminosity Physics pile-up (simulated, with big uncertainties) low luminosity 30 MeV high luminosity 100 MeV Total contribution low luminosity 155 MeV high luminosity 210 MeV Measured at 2003 testbeam (over 3x3) = 180 MeV Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Energy resolution: c leakage (front, rear, dead material) CMS full shower simulation < 0.2 % system instabilities (designed to be at the permill level t 3 tcal ) temperature stabilization at 0.05 ˚C (dLY/dT = -1.9 %/˚C @ 18˚C ; dM/dT ~ -2.4 %/˚C) APD bias stable at 20 mV (dM/dV = 3%/V) It is the limiting factor at high energies. Need to keep c 0.5 % : light collection uniformity intercalibration by monitor and physics signals at 0.5 % including the radiation damage effect Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Radiation Damage [cm] EE Total dose after 10 years (5x105 pb-1) EB Dose rates [Gy/h] in ECAL at L=1034cm-2s-1 0.15 [Gy] The only damage is produced by e.m. radiation on crystal transparency (colour centers) The damage level depends on the dose rate An equilibrium is reached after a small dose Partial damage recovery in few hours Loss in extracted light (~ 3 % for 0.15 G y/h) can be monitored with a laser 1 2 3 4 5 6 7 8 i n t a l f e r d o w v g h ( m ) T(%) Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Laser monitoring system Need to monitor crystal transmission continously to follow damage and recovery during LHC cycles inject laser light in each crystal follow signals from beam and laser determine slope laser versus beam check universality of the slope Two laser systems: at 440/495 nm (blue, green) and 700/800 nm (red, infrared) with diagnostics on wavelength, pulse shape and intensity Laser signals injected in the single crystals and in four pin diodes (which have 0.1 % stability) Full calorimeter measurement foreseen every ~ 1 hour by using LHC gaps (3 ms every 89 ms) Able to keep inter-calibration coefficients within 0.4 % over two months Can recover calibration coefficients after power cuts Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Scheme of the laser Absolute linearity study of PN readout electronics Relative linearity study of APD readout electronics with respect to PN diodes Beam : Consistency check of the method s =0.14% Stability of the monitoring system achieved at 0.1 % level by normalizing to the PN diodes Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Read out chain/old 40 MHz clock High dynamic range (50 MeV 2 TeV) Radiation hardness Low noise Upper-Level VME Readout Card (in Counting Room) Pipeline S To DAQ Digital Trigger S Energy ® Light Current Voltage Bits PbWO4 Crystal APD VPT Floating-Point Preamplifier Fiber Readout ADC On-detector Light-to-Light readout Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Read out chain/new Major revision imposed by budget (cost of optical links) and possible thanks to 0.25 mm CMOS rad hard technology trigger and data storage on detector read out of data only on receipt of a L1 trigger (3 ms) 3 Gigabit Optical Links each trigger tower (25 crystals) Reduction of about a factor 8 in the number of data links Simplification of off-detector electronics Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Electronics - pictures MGPA card trigger tower LVR card Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Cooling system Power ~ 2.5 W/channel !! Cooling requirements: DT ~ 0.05 0C over 3 x calibration time DT ~ 0.2 0C uniformity within a supermodule 2.5 W/channel to be taken directly from electronics chips (to avoid convection) Need a good thermal coupling between the electronics and the cooling water Need to minimize the heat flow to the APD Stainless steel cooling pipes embedded in Al, thermally coupled with gap pad / gap filler (Berquist) to the electronics components Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Cooling system Mean = 0.043 K RMS = 0.013 K Test to validate the cooling system (switch off and on electronics) successful trigger tower concept cooling bar production Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Cooling : pictures Housing with gap pad Mother Board on the Grid Mother Board installation Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Cooling : pictures VFE and LVR cards on the cooling bars Trigger tower on the cooling bars Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Trigger tower Trigger GOH FENIX FPGA VFE QPLL CCU Readout GOH LVR Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN Testbeams 2002 testbeam : 100 channels with old electronics, cooling and monitoring systems not final Calibration table + hodoscope Behaviour of APDs, HV, LV, VFE electronics Universality of irradiation behaviour Check of pre-calibration from LAB measurements 2003 testbeam : 50 channels with the new electronics, final cooling and laser monitoring noise level, auto-gain switching stability of cooling and monitoring systems 2004 testbeam : 1 supermodule Final system test and complete calibration Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Testbeam: laser monitoring Dispersion of a for 19 crystals Use of same a coefficient for all crystals possible ! s/m = 6.3% on top of a 5 -10 % effect signal from Laser signal from Beam (e-) a slope : a = 1.55 Follow up of intercalibration during irradiation with 0.4 % accuracy Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Testbeam: precalibration M.Paganoni, Univ. Milano Bicocca and INFN Comparison of Light Yield (LAB - BEAM)/BEAM 97 crystals : RMS = 4.57 % Sigma = 4.40 % Obtain crystal intercalibration from measurements of light yield in the LAB Put crystals on electron BEAM and check intercalibration Compare LAB and BEAM LY/LYref Intercalibration can be obtained from LAB with a precision of 4.5 % ! Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
Testbeam: energy resolution Good energy resolution achievable with 0.25 mm electronics after correlated noise subtraction Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN The calibration at LHC Goal : follow the evolution of intercalibration at few permill level (absolute energy scale from Z -> ee) Starting point : calibration on beam for as many modules as possible (2% accuracy) and laboratory measurements (Light Yield,transparency, APD multiplication and electronics gain) for the others (5% accuracy) Calibrate every month with physical events (W e n ; Z ee), at 0.3% level Follow up of the radiation damage with the laser monitoring (0.4% precision) Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
M.Paganoni, Univ. Milano Bicocca and INFN The achieved goals Increase of the light yield, while keeping a fast response and longitudinal response uniformity Development of a solid state photodetector with gain (APD) to be operated in a high magnetic field Radiation hardness of crystals and electronics Fast growing of crystals and mechanical assembly Design and validation of the cooling system and the monitoring system Test of the pre-calibration procedure and of the follow-up of the radiation damage Validation of the new electronics Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN
….and the next challenges End the production of 75000 crystals and check their quality Assemble the whole detector Calibrate on beam as many supermodules as possible Monitor the calorimeter during the precalibration and all the data taking at the few permill level Vienna, VCI04, 20/2/2004 M.Paganoni, Univ. Milano Bicocca and INFN