Electrons and Photons at CMS and Tests with

Electrons and Photons at CMS and Tests with DØ Data Yuri Gershtein 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven

Outline The problem precision calorimetry with dead material in magnetic field CMS and DØ detectors / algorithms “CMS-like” algorithm performance in DØ data Z, J/ , What is different between CMS and DØ new challenges Outlook 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 2

Electrons and Photons at CMS Lead Tungstate crystals Excellent stand-alone resolution Biggest challenge is the amount of material in front of the ECAL CMS 2/3/2005 CMS Yuri Gershtein - Tev 4 LHC Brookhaven 3

Calorimeter: Dense and Finely Segmented 61200 barrel crystals No longitudinal segmentation 14648 endcap crystals Transverse segmentation ~ 0. 02 x 0. 02 Four longitudinal layers Transverse segmentation ~ 0. 05 x 0. 05 in shower max ~ 0. 10 x 0. 10 other layers 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 4

Tracking: Few Precise Measurements Radius ~ 55 cm, length ~ 260 cm 16 layers of Scintillating Fibers (8 axial + 8 stereo) 4 to 8 silicon layers (mostly double-sided) Radius ~ 110 cm, Length ~ 540 cm h~1. 7 6 layers TOB h~2. 4 4 layers TIB 3 disks TID 2/3/2005 9 disks TEC Yuri Gershtein - Tev 4 LHC Brookhaven 5

Algorithms DØ CMS: find bumps in calorimeter cluster the bumps in barrel the window size ~ 0. 8 0. 06 2/3/2005 “Nearest Neighbor” algorithm (Cell. NN) find deposits consistent with single particle in each layer Parameters are tuned to avoid EM shower “splitting” Cell energies are shared between clusters according to EM shower size Yuri Gershtein - Tev 4 LHC Brookhaven 6

DØ: Z ee with Cell. NN Two highest Et Cell. NN clusters + two tracks Mee (Ge. V) Subset of events which has at least one Cell. NN cluster near the electron candidate ( R < 0. 2) 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven Mee (Ge. V) 7

DØ: Z ee with Cell. NN After correction Before correction for extra clusters Mee (Ge. V) internal + external brems no “ -spray” pattern 2/3/2005 • bend angle is small Yuri Gershtein - Tev 4 LHC Brookhaven 8

DØ: J/ and A little less dramatic improvement – lower p. T and poorer resolution Cell. NN threshold is 1. 5 Ge. V – needs to be lowered all before / after Mee (Ge. V) 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 9

When Quantity Transforms to Quality Many experiments dealt with recovery of energy lost to Bremsstrahlung and dead material New combination: huge amount of material in 4 T field excellent intrinsic calorimeter resolution Effects that were “small” in other experiments are not small any more CMS Tracker material – up to 1. 5 X 0 DØ Preshower detector – 2. 0 X 0 CMS curler p. T ~ 0. 8 Ge. V DØ curler p. T ~ 0. 2 Ge. V 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 10

How Much of Electron Energy Reaches Calorimeter? Run GEANT simulation of CMS Keep track of all Brems and conversions Propagate all particles to calorimeter face bremsstrahlung photons convert and electrons from conversion curl up So far look in barrel only (Occupancy in endcap is much clusterable larger problem) does not reach ECAL 2/3/2005 too far from the main cluster Yuri Gershtein - Tev 4 LHC Brookhaven 11

CMS Barrel at calorimer face electrons with ET 80 Ge. V flat in rapidity and azimuth in 0. 8 x 0. 06 window How to estimate resolution: determine the window that has 95% of events, calculate RMS in this window ET ET If one knows amount of tracker material one can correct for average lost energy pseudorapidity 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 12

“Small” Effect Resolution before dependent correction Resolution after dependent corrections Intrinsic resolution of the calorimeter / ET 20 2/3/2005 40 Yuri Gershtein - Tev 4 LHC Brookhaven ET 13

Photons v. s. Electrons Unlike electrons, photons penetrate through material completely intact until the first conversion photons mean = 39. 86 electrons mean = 39. 44 95% Even with perfect calorimeter ~ 1% average difference between electron and photon energy scales @ 40 Ge. V 1. 5% difference if use only unconverted photons TDR figure for unconverted photon energy resolution is 0. 9% at 35 Ge. V If calibration is done with electrons from W and Z -> need to know material in order to derive photon energy scale 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 14

Lessons from Tevatron Material is hard to get right especially it’s spatial distribution CDF was not able to use electron track momentum for W mass measurement in Run I DØ Run II material was initially underestimated by almost a factor of 2 Engineering drawings do not correspond to “as built” detector Note: the distribution of the material in azimuth is very non-uniform DATA 1000 k 2/3/2005 MC 60 k Yuri Gershtein - Tev 4 LHC Brookhaven DØ can use -symmetry for calorimeter inter-calibration because calorimeter resolution is worse than in CMS and there is not as much material 15

Measurement of Material Use well-measured resonances, like J/ , to tune MIP momentum measurement in tracker gives “average” material Photon conversions give full 3 -D material distribution that can be fed back to simulation Challenge is to determine reconstruction efficiency Some handles from KS – known lifetime Some handles from symmetric / asymmetric conversions Very delicate measurement! The systematic and the tools are quite similar for CMS and DØ (same tracker philosophy - few precise measurements on a track) Studies in DØ are still ongoing, >3 years into the run… 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 16

Are “Lost” Brems Unrecoverable? - (MC) Consider a simplified situation: only one brem per electron occurring at fixed radius Even if the brem is lost energy and position measurement in ECAL and initial direction measurement from pixel detectors can be used to calculate the energy of the brem and “recover” it Even the crudest combination already shows correlation: ET 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 17

Full Simulation (ECAL) - (pixel) Same correlation is clearly seen in the full simulation ( is corrected by bend angle expected from calorimeter energy measurement) Super-cluster ET 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 18

Interplay of Tracker and Calorimeter an algorithm could be devised that would make a global fit to an electron allow for track p. T changes identify brems takes full advantage of the calorimeter – track correlations 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 19

Summary The DØ data qualitatively confirms that CMS-like Bremsstrahlung recovery works The problem of Bremsstrahlung recovery at CMS however is not just about finding all clusters in the calorimeter, it’s about correcting for the energy that does not reach the calorimeter at all Photon and electron calibrations are therefore substantially different One of the biggest challenges is making a precise measurement of the material before the calorimeter a tricky measurement especially hard with a tracker with very few (~10 -15) layers DØ methods/tools for that might be directly applicable to CMS 2/3/2005 Yuri Gershtein - Tev 4 LHC Brookhaven 20