An Overview of the CMS ECAL with Applications to HEP Analysis - PowerPoint PPT Presentation

An Overview of the CMS ECAL with Applications to HEP Analysis Daniel Klein Thursday Pizza Lecture 11/14/2013

Outline I. Motivation 1. What is an ECAL? 2. Design requirements for CMS ECAL II. Design 1. Materials 2. Crystal geometry 3. Large-scale geometry III. Measurement 1. Superclustering 2. Triggers 3. Particle reconstruction and selection 14-Nov-2013 2

Motivation 14-Nov-2013 3

What is an ECAL? What does it do? ● Stands for Electromagnetic CALorimeter ● Used to measure the energy of electrons/positrons and photons, and (indirectly) their parent particles ● Help with identification of EM particles (more on this later) ● Help determine (rough) positions of EM particles, in conjunction with tracker 14-Nov-2013 4

Some physics goals that influenced CMS ECAL design ● Higgs search – H → γγ dominant decay mode for 114 GeV < m H < 130 GeV – H → ZZ → 4ℓ the “mode of choice” for 2m Z < m H < 600 GeV ● SUSY searches – GMSB: LSP → + γ (expect lots of hard photons) – / → γ + jets ● New vector bosons – Z' → ee ● Lots and lots of standard model physics 14-Nov-2013 5

Technical Requirements From TDR: Summary of ECAL requirements in order to meet LHC physics program goals: ● “Good” electromagnetic energy resolution ● ee and γγ mass resolution of ~1% at 100 GeV ● Coverage out to |η| = 2.5 ● Measurement of γ direction, or PV localization ● Rejection of π 0 ● Efficient photon and lepton isolation at high luminosity 14-Nov-2013 6

Design 14-Nov-2013 7

Materials ● Primary detection material: lead-tungstate crystals (PbWO 4 ) – Radiation length X 0 = 0.89 cm Recall: – Moliere radius R M = 2.2 cm – Fast: 80% of light emitted within 25ns. Comparable to bunch-crossing time. – Radiation-hard – up to 10 Mrad – Emit blue-green scintillation light peaking at ~420 nm ● Photodetectors ● Endcap also has preshower detector – Stuck onto the back of each crystal – Sits just inside endcap crystal array – Barrel: silicon avalanche photodiodes – Sampling calorimeter (APDs) – (Lead “radiator” + silicon strip sensors) * – Endcap: vacuum phototriodes (VPTs) 2 layers 14-Nov-2013 8

Crystal geometry/resolution ● Reminder: ● Crystals shaped like truncated pyramids ● Barrel section: – Rad. length X 0 = 8.9 mm – Made of 61,200 crystals – Front face: 22x22mm = 1x1 R M ~ 1°x1° – Moliere radius R M = 22 mm – Length: 230mm = 25.8 X 0 – Most energy (~94%) from a single particle will be contained in 3x3 crystals ● Endcap section: – 2x endcaps, containing 7324 crystals each – Front face: 28.6x28.6mm = 1.3x1.3 R M – Length: 220mm = 24.7 X 0 – Most energy from a particle will be contained in 3x3 crystals 14-Nov-2013 9

Energy Resolution (In case you're not sick of this plot yet...) → Comes from electron test-beam studies on a supermodule. 14-Nov-2013 10

Large-scale geometry: Barrel ● Range: 0 ≤ |η| ≤ 1.479 ● Inner radius: 1.29 m ● 61,200 crystals = 360 around * 170 lengthwise ● 5x2 crystals in a “submodule” – Each submodule matches up with a trigger tower in η and φ ● Submodules arranged into modules ● 4 modules (85x20 crystals) in one “supermodule” – Each covers ½ the length in η and 20° in φ (36 total) ● Crystal axes point 3° away from nominal interaction point 14-Nov-2013 11

Large-scale geometry: Endcaps ● Range: 1.479 ≤ |η| ≤ 3.0 ● Set back 3.14 m from nominal interaction point ● Each endcap made of two “Dees,” 3662 crystals per dee ● Crystals are arranged in 5x5 “supercrystals” – Each dee holds 138 supercrystals and 18 partial supercrystals ● Supercrystals arranged in an x-y grid, NOT an η-φ grid. ● Crystal axes point to a spot 1.3 m past the nominal interaction point 14-Nov-2013 12

Particle Reconstruction 14-Nov-2013 13

ECAL superclustering ● Photon conversion and electron bremsstrahlung cause shower to be spread out in φ direction. – Form “superclusters” - clusters of clusters, with some spread in φ ● Hybrid algorithm: start with a “bar” 3-5 crystals wide in η, then search dynamically in φ for more deposits – Works well for high-energy electrons in barrel ● Island algorithm: start with one crystal, then keep adding adjacent crystals with energy deposits until you form a cluster – Add nearby clusters (within a narrow η window, broader φ window) to form a supercluster – Works well when small, isolated clusters are needed ● Use log(energy)-weighted averaging to find center of a cluster 14-Nov-2013 14

Supercluster examples Probably island algorithm Probably hybrid algorithm 14-Nov-2013 15

Triggers ● Level 1 trigger: E T threshold, applied to superclusters that match in η and φ with a trigger tower – 50% efficiency levels: single isolated: 23 GeV, double isolated: 12 GeV, double non-isolated: 19 GeV – Isolation determined from HCAL and tracker ● High-level trigger (HLT) selection has three sub-levels: – Level 2: an E T cut on ECAL superclusters – Level 2.5: Look for pixel hits in tracker consistent with an electron (positron) hypothesis – Level 3: If passing level 2.5, use full tracker info (including tracker isolation) to attempt to match tracks to ECAL deposit ● If a deposit doesn't pass the level 2.5 trigger, it can still be used as a photon candidate ● Object-specific HLT cuts: 14-Nov-2013 16

Photon Reco & Selection ● Energy is a sum over 5x5 cluster, or hybrid supercluster (EB), or island supercluster (EE) ● 3 tracker-based isolation variables used, based on sum pT, angle, or number of tracks within some cone size of ECAL cluster – Used to reject photons from charged π or k ● 4 ECAL isolation variables used, based on energy deposited in a certain cone size around supercluster, or on R9 (E 3x3 / E supercluster ) – Used to reject photons from π 0 ● HCAL isolation based on simple sum of HCAL E T in a cone around ECAL supercluster – Used to reject photons from jets – H/E variable shows worse performance than simple sums in HCAL ● Variables from multiple subsystems are also combined using neural networks ● Also use tracks to reject photons that converted 14-Nov-2013 17

Electron (positron) Reco & Selection ● Bremsstrahlung spreads out electron energy in φ – Brem photons can even convert in tracker – Electron energy best measured using superclusters, not NxN windows ● Electron ID makes heavy use of tracker information, including isolation, E/p, primary vertex reconstruction, etc. – Another slideshow unto itself (Liam) ● Shower shape variables used in electron ID include: σ ηη , Σ 9 /Σ 25 ● HCAL isolation used to reject electron candidates coming from jets 14-Nov-2013 18

Example (ECAL-based) cuts from CMS2 NtupleMacros/CORE/ NtupleMacros/CORE/ electronSelections.cc photonSelections.cc ● electronIsolation_ECAL_rel_v1 ● cms2.photons_ecalIso03 < < 0.20 [pt-dependent threshold] ● Transition region veto (reject ● Barrel-only (η < 1.479) 1.442 < η < 1.556) ● cms2.photons_hOverE < 0.05 ● cms2.els_hOverE < 0.15 ● cms2.photons_sigmaIEtaIEta < ● cms2.els_eOverPIn > 0.95 0.013 14-Nov-2013 19

Summary ● CMS requires an efficient, high-precision electromagnetic calorimeter ● This requirement was met by designing an ECAL made mostly of lead-tungstate crystals, with scintillation light read out by photodiodes/triodes – Crystals have short radiation length and Moliere radius, allowing fine resolution in eta and phi ● Energy deposits are collected into (super)clusters, the basic blocks of energy measurement ● Measurements from other detector subsystems aid in ID and selection of electrons and photons 14-Nov-2013 20