Beijing, China May 5, 7 Performance of the CMS electromagnetic calorimeter during the LHC Run II Chia Ming, Kuo National Central University, Taiwan on behalf of the CMS Collaboration
ECAL is crucial in CMS physics analysis New physics searches and Standard Model precision measurements with photons and electrons S/(S+B) Weighted Events / GeV 8 6 4 8 6 4 4 CMS H γγ Preliminary H γγ H ZZ 4l Z ee 5.9 fb ( TeV) All categories S/(S+B) weighted Data S+B fit B component 6 B component subtracted 4 5 6 7 8 m γγ (GeV) Events / 4 GeV Data H(5) qq ZZ, Zγ* 8 gg ZZ, Zγ* Z+X 6 4 CMS Preliminary 8 4 5 5.9 fb ( TeV) 7 9 m 4l (GeV) Events / GeV 5 Data 4 4 + - γ*/z e e tt, tw, WW, WZ, ZZ, ττ Jets.4 fb ( TeV) CMS Preliminary 8 m(ee) [GeV] high energy resolution position measurement a wide range of ET high energy resolution energy linearity TIPP 7 7/5/5
CMS Electromagnetic Calorimeter Preshower (ES) subdetector η-coverage Endcap (EE) Barrel (EB) read-out channels Compact enough to fit inside the.8t superconducting solenoid X homogeneous, hermetic, compact, fine-grain PbWO 4 crystal calorimeter density of 8. g/cm short radiation length.89 cm small Moliere radius. cm fast light emission : ~8% in ~5 ns refractive index =. light yield spread among crystals % (RMS) from beam test strengths : precise e/γ energy and position measurements good timing resolution fast and efficient readout for online selection (DAQ and trigger) TIPP 7 7/5/5
ECAL alignment position reconstructed from energy deposit exploiting ECAL high granularity electron identification : prompt vs fake electrons required matched measurements between ECAL and tracker is better than. radians in Φ and.4 units in η measurement of photon direction : H γγ procedure based on matching position reconstructed by tracker and ECAL with Z ee events Number of events 6 4 CMS Preliminary 6 8 6 4 RMS DATA after alignment - =. - RMS DATA before alignment =. EB.5 fb ( TeV) 8 6 4 4 6 8 MC DATA after alignment DATA before alignment η Number of events..8.6.4. CMS Preliminary 6 RMS DATA after alignment - RMS DATA before alignment =. EE η < 5 5 5 5 TIPP 7 4 7/5/5 =..5 fb ( TeV) - 8 6 4 4 6 8 MC DATA after alignment DATA before alignment η Number of events 9 8 7 6 5 4 CMS Preliminary 6 RMS DATA after alignment RMS DATA before alignment EE η < = 5. = 5.6.5 fb ( TeV) - - MC DATA after alignment DATA before alignment φ
Energy reconstruction Electrons and photons deposit energy over several crystals (7% in one, 97% in a array), spread in Φ, collected by clustering algorithms Laser Monitoring Inter-calibration Global Scale Cluster Corrections Pulse Amplitude CMS ECAL energy resolution : uniformity and stability resolution required in situ <.5% in barrel, % energy resolution achieved in Run-I and Run-II for unconverted/ late-converting photons TIPP 7 5 7/5/5
Pulse reconstruction Calibrated/Pedestal-Subtracted Energy (GeV) When LHC runs with 5-ns bunch-spacing, the level of out-of-time (OOT) pile-up increases to mitigate this effect Multifit algorithm : pulse shape is modeled as a sum of one intime pulse pluses OOT pulses χ = 8 7 6 5 4 CMS simulation, s= TeV " i= Observed signal Total pulse In-time pulse Out-of-time pulses ( M j= A j p ij S i ) σ S i digitized signals are recorded and used for pulse reconstruction PU=/BX, 5 ns 4 5 6 7 8 9 Time sample Sample BX =, -4,, +, etc. fitted up to 9 OOT pulses (one per time sample) minimize χ distribution for best description of the in-time amplitude LHC isolated bunches are used to extract the pulse shapes (binned templates) periodically baseline and electronic noise periodically measured from dedicated runs and used in the covariance matrix Minimization using non-negative least-squares : fast enough to be used both offline and online TIPP 7 6 7/5/5
Crystal response monitoring tracker coverage precision physics high η ( η >.5) jet physics Steady recovery during shutdowns and inter-fills In the regions close to beam pipe, not fully recovered ECAL radiation-induced effects, heavily η dependent crystal transparency changes VPT photocathode aging with accumulated charge channel response is constantly monitored with a laser system injecting light in every ECAL crystal calibration point per channel every 4 mins corrections obtained and applied in ~48 hours for prompt reconstruction TIPP 7 7 7/5/5
Validation of response monitoring Response stability after corrections validated with physics signals: the stability of π invariant mass E/p relative scale of isolated electrons from W decays mass π normalized..98 CMS preliminary 6 ECAL Barrel With LM correction mean =.. r.m.s =.7 %.98 Relative E/p scale..98 CMS 5 Preliminary.4 with laser monitoring correction s = TeV, L =.5 fb without laser monitoring correction Mean RMS.57 Mean.94 RMS.6944.96.96.96.94.9 Without LM correction FILL 4958 : 8 May 6 4h 5h 6h 7h 8h 9h h h time.94 mean =.94 r.m.s =. %.9 5 5 5 π mass history in barrel in fill <signal loss> ~% RMS after laser monitoring corrections : ~.7% very fast monitoring : point/8-min.94.9 ECAL Barrel 7/9 4/9 / 8/ 5/ / 9/ date (day/month) Electrons E/p history in barrel in 5 <signal loss> ~6% 4 6 RMS after laser monitoring corrections : ~.5% TIPP 7 8 7/5/5
Inter-calibration (IC) Equalizes the response of each single crystal to the deposited energy pairs /.4 GeV/c γγ 5 5 5 5 CMS Preliminary. fb ( TeV) 4 Data Signal + background Background only ECAL Barrel crystal.8...4.6.8. γγ invariant mass (GeV/c ) Inter-Calibration Precision.4... CMS Preliminary - ECAL barrel φ-symmetry electron π / η combination.5 crystal η IC residual miscalibration (%) CMS Preliminary 4.5.5.5.5 ECAL barrel +5 phi-symmetry π electrons combination.5.6fb ( TeV) crystal η Method Description Run-I precision Φ-symmetry π /η γγ Energy flux around Φ rings (constant η) should be uniform - IC corrects for non-uniformity In a Φ ring, use IC to improve Mγγ resolution for π and η resonances -% in EB -5% in EE.5% in EB % in EE ( η< ) electron E/p Compare isolated electron energy from ECAL and tracker, compute IC to correct discrepancies.5% in EB % in EE TIPP 7 9 7/5/5
η scale and absolute calibration the η dependence of the energy reconstruction and its absolute scale are calibrated with electrons from Z decays Z peaks in a single η-ring are used to correct the relative scale between different η-rings The Z peak is used again to fix the overall absolute calibration, matching data to a detailed simulation of the detector EB-EB.8 T EB-EB T EB-EE.8 T EB-EE T separate absolute calibrations for.8 and T data The inclusion of T data improved the search sensitivity of X γγ by ~% TIPP 7 7/5/5
Clustering and corrections Events / GeV 6 5 4 CMS Preliminary s = TeV, L =. fb ECAL Barrel-Barrel EB - EB SuperCluster corrected SuperCluster raw E E E 5x5 crystals large amount of material before ECAL dynamic clustering algorithm recovers energy radiated upstream of ECAL via bremsstrahlung or conversions super-cluster (SC) of clusters along Φ (bending direction) soft conversion legs/brem may not be included in SC in the endcaps, preshower energy is also considered additional energy from pileup contaminates the shower the energy of supercluster is corrected using a multivariate approach that maximally exploits the information of the events tuned on MC, validated on data Reconstructed Z mass in data with different levels of energy reconstruction and corrections Events / GeV 5 6 7 8 9 + - CMS Preliminary s = TeV, L =. fb 9 8 7 6 5 4 ECAL Endcap-Endcap EE - EE m(e e ) [GeV] 5 6 7 8 9 + - SuperCluster corrected SuperCluster raw+es SuperCluster raw E E E E 5x5 crystals m(e e ) [GeV] TIPP 7 7/5/5
Energy and mass resolution derive electron energy resolution from Z ee peak width improvement from prompt to refined conditions for η <, precision at the level of Run-I simulation tuned to match resolution observed in data Events /.5 GeV 5 4 CMS Preliminary s = TeV, L =. fb ECAL Barrel-Barrel EB-EB Data Z+jets 75 8 85 9 95 5 + - m(e e ) [GeV] Events /.5 GeV 5 5 5 5 CMS Preliminary s = TeV, L =. fb ECAL Not Barrel-Barrel EB-EE EE-EE Data Z+jets 75 8 85 9 95 5 + - m(e e ) [GeV] / E σ E.8.7.6 low bremsstrahlung electrons Simulation ( fb precision), R 9 Prompt reconstruction, R 9.94.94 Winter5-6 re-reconstruction, R 9.5 fb.94 ( TeV) CMS Preliminary / E σ E.8.7.6 high bremsstrahlung electrons Simulation ( fb precision), R 9 Prompt reconstruction, R 9 <.94 <.94 Winter5-6 re-reconstruction, R 9 <.94.5 fb ( TeV) CMS Preliminary.5.4....5.5.5 Supercluster η.5.4....5.5.5 Supercluster η TIPP 7 7/5/5
Estimation of single e/γ resolution per-electron or per-photon resolution used to build a per-event mass resolution (σ m /m), utilized to make optimal use of the highest resolution events H ZZ 4l : per-event mass resolution used as a variable in the fit for mass measurement validated in data with fits to Z ee by comparing the predicted σ ml /m l H γγ : used to classify in several untagged categories for m γγ fit - lnl H ZZ 4l mass likelihood observed vs predicted σm H γγ best category 8 7 6 5 4 CMS Preliminary m' 4l m' 4l, D' mass kin m' 4l, D' mass, Dbkg kin m' 4l, D' mass, Dbkg 5.9 fb ( TeV) (stat. only) Measured σ ml /m l.5.4... CMS Preliminary - Z e + e - Z µ + µ Data Data 5.9 fb ( TeV) Events / (.5 GeV ) 7 6 5 4 CMS Simulation Preliminary H γγ Simulation Parametric model σ eff =. GeV FWHM =.9 GeV TeV Untagged σm/mγγ ~ % 4.5 5 5.5 6 m H (GeV)....4.5 Predicted σ ml /m l 5 5 5 5 4 m γγ (GeV) TIPP 7 7/5/5
Energy resolution for high energy photons recalibration also has an important impact on high energy photons saturation effects of electronics corrected with multivariate approach single channel saturation in barrel : E ~.6 TeV impact on energy scale < % residual non-linearity checked with boosted Z ee : <.5% (.7%) for photons up to 5 GeV in the barrel (endcap).9 fb ( TeV) Fraction of events arbitrary unit CMS Simulation.5..5. -4 m m Γ = 5 GeV =.4 Prompt-reco Re-reco ( TeV,.8T) EBEE Fraction of events arbitrary unit.5..5..5 CMS Simulation m -4 m Γ = 5 GeV =.4 Prompt-reco Re-reco ( TeV,.8T) EBEB Events / GeV CMS EBEB Data Fit model ± s.d. ± s.d..5 44 46 48 5 5 54 56 (GeV) m γγ..5 44 46 48 5 5 54 56 (GeV) m γγ (Data-fit)/σ stat - 5 5 m γγ (GeV) TIPP 7 4 7/5/5
Conclusions The CMS electromagnetic calorimeter performs well during LHC Run-II and plays a crucial role in physics beyond SM searches and precision measurements including Higgs physics Continuous developments and understandings of the detector details new amplitude reconstruction algorithm in place to cope with ~4 pileup interactions ready for even higher values expected in 7 in barrel, % energy resolution achieved in Run-I and Run-II for unconverted/late-converting photons re-calibration with 6 data is ongoing stay tuned with m H measurement in γγ final state with Run-II dataset TIPP 7 5 7/5/5