CMS (ECAL) precision timing T. Tabarelli de Fatis Università di Milano Bicocca and INFN Recent studies and technical choices are not all public Not the latest snapshot in these slides 1
Bold aspects of the CMS upgrade for HL-LHC High granularity endcap calorimeters Energy, position and time mapping of showers Track information in HW event selection Tracker with higher granularity and extended acceptance ( η <4) } Goal: be as efficient, and with low background/fake rate, at 2 pileup as we are today and with an extended acceptance 2 Precision timing of all objects to combat pileup New electronics in ECAL barrel Dedicated layers for single charged track timing just outside the Tracker
recisiontimingfinalreport#draq_4_1_217_mar_16th((( Time-aware reconstruction t (ns) If beam-spot sliced in successive O(3) ps time exposures, effective pileup reduced by a factor 4-5: ~15% merged vertices reduced to 2% Phase-I track purity of vertices recovered Simulated Vertices 3D Reconstructed Vertices.6 4D Reconstruction Vertices 4D Tracks.4 vertices time.2 -.2 z -.4-15 -1-5 5 Luminous region trms ~ 18 ps zrms ~ 4.6 cm VBF H!ττ in 2 pp collisions 3 VBF(H(#(ττ(in(2(p,p(collisions(
Track-vertex association with track timing Track-PV association pileup fraction 1.9.8.7.6.5.4.3.2.1 CMS Simulation preliminary tt event tracks p >.9 GeV T HL-LHC BS, 3D vtx, PU=14 HL-LHC BS, 3D vtx, PU=2 HL-LHC BS, 4D vtx, PU=14 HL-LHC BS, 4D vtx, PU=2 13 TeV 3D RECO 4D RECO LHC HL-LHC 14PU 2PU.2.4.6.8 1 1.2 1.4 1.6 1.8 2 Density (events/mm) } With timing, effective vertex density down to LHC level! 1. Extend performance at 2 PU 2. Strengthen reconstruction at 14 PU 3. Provide robustness against adjustment of luminosity scenarios } Recovery from performance degradation in several observables 4
Example: track isolation efficiencies w/ timing w/ timing Muons no timing Taus no timing } Acceptance gain in searches and precision measurements } [ Curves are for constant background reject power ] } Performance benefits also in, missing E T resolution, pileup jet rejection, 5
Photon timing and association with vertices } Association with vertex requires t information [ i.e. charged track timing ] } Single photon: } early collision + long TOF = late collision + short TOF t (ns) } Two-photons } triangulation: vertex space- time from photon timing } For small rapidity gap triangulation breaks down } Vertex t to resolve ambiguities 6 Matching Neutrals to 4D Vertices 2.6.4.2.2.4 CMS Simulation <µ> = 2 Simulated Vertices 3D Reconstructed Vertices 4D Reconstructed Vertices 4D Tracks Leading Photon Vertex Hypotheses η = -1.26 Sub-leading Photon Vertex Hypotheses η = -.66 Δη(γγ) ~.5 t = t + TOF Few vertices shown to ease eye analysis Hàγγ event (3 ps resolution) 6 4 2 2 4 z (cm) Above is a space-time diagram demonstrating the inability of close-by photon resolve a vertex alone, using a H γγ decay as illustration. The reconstructed
Elements of the timing upgrade of CMS } Calorimeter upgrades: } Precision timing (~3 ps) of high energy photons in ECAL, photons and high energy hadrons in HGCal } Investigating low energy hadrons in HGCal } Additional (thin) timing layers } MIP timing with 3 ps precision and almost full efficiency } Just outside the tracker: } Acceptance: η <3. and p T >.7 GeV 7
Timing in the CMS upgraded calorimeters } HGCal (left): Si + Cu/W or Pb/Steel: 28 layers.5 and 1 cm 2 cell size 5 ps / cell [at least for >2 MIPs signals] } ECAL (right): PbWO 4 + APDs New electronics + 16 MHz sampling Cluster time resolution 5 GeV Single cell ~5 ps 3 ps } Investigating low energy hadrons in HGCal 8 ~3 GeV
Timing tests of the HGCal Si sensors Tim ingresolutionvsam plitude Time resolution improves with S/N Constantterm:14ps,forS>2mips Tim ingtestwithspecialfastreadout Timing test with 3 m layer Fastreadout 16 psfor32gev electrons CERN beam tests in November 216 Upto25GeVelectrons Analysisongoing 9
Barrel MIP timing detector } LYSO crystals + SiPM embedded in the Tracker tube } } Ready before TK integration Maintain performance at radiation levels of 2x114 neq/cm2 Tracker Support Tube ~4 m2 Modules (16x4 crystals) Concentrator Card 4 FE Boards Electronics readout: Adapt TOFPET2 ASIC 1 TOFPET CHIP
Barrel sensors (LYSO + SiPMs) } Nominal geometry: 12 x12 mm 2 (~ 3 mm thick) + 4x4 mm 2 SiPMs } Production-like geometry qualified in test beams } Good radiation hardness of production-ready SiPMs } Operate SiPMs at ~ 35 C (limit self-heating and dark rate) Test beam σ Δt / 2 = 21 ps SiPM Crystal 11
Radiation hardness: SiPMs } Radiation fluence is a fierce foe } SiPMs self-heating ~ 3 mw/channel } Dark count rate (DCR) a few 1 GHz/mm at -35 C } LYSO scintillation to boost signal above dark counts noise } Optimal SiPMs require balance between photon detection efficiecy (PDE) and DCR to maintain 3 ps at the end of life 12 LYSO is radiation hard at the required level
Endcap MIP timing detector } Single layer of low gain silicon detectors just outside TK } Ready at the end of TK or HGCal integration } Maintain performance at radiation level 2x1 15 n eq /cm 2 Neutron Moderator Endcap Calo ~2 x 5 m 2 η TK / HGCal nose 1.6 1.1 x 1 14 2. 2.1 x 1 14 2.5 4.8 x 1 14 3. 1. x 1 14 13 Fluence [n eq /cm 2 ] at 3 fb -1 Single layer OK here Requires further R&D (or multiple layers) η = 1.5 3. HGCal
Silicon detectors with gain } Different gain / E-field geometries under study (RD5) 14 Additional doping layer to achieve gain Low Gain Avalanche Diodes (LGADs) Gain O(1) - Three suppliers (FBK, CNM, HPK) Ultra Fast Si Detectors Drift region ~ 5 μm to limit Landau fluctuations Small pads to limit capacitance Details on the technology in M.Obertino s talk 4 μm avalanche region Deep Depleted APDs Gain O(5) - one supplier (RMD) 15 ps on 1 cm 2 pads Not yet fully qualified at CMS radiation levels [S.White, at Frontier Detectors etc., Elba, (Italy) 215 ]
Endcap sensors (ultra-fast Si detectors) } Nominal geometry: 4.8 x 9.6 cm 2 modules with 1x3 mm 2 pads } 16 ASICs bump-bonded to sensors (ASIC being developed) } 3:1 ganging in the TDC at small η (3x3 mm 2 granularity) } Single pads shown to have σ t 5 ps up to 1 15 n eq /cm 2 } Compensate gain loss with bulk gain (higher external bias) Inner rings 24 channels 4 channels Outer rings 4 channels ROC for small pixels 1 channel 8 channels is 1x3 mm 2 J. Lange, JINST, P53 (212) 3e14-6 C 1e15-15 C ROC for large pixels 1 channel is 3x3 mm 2 3e14-2 C 15 Details on the technology and R&D prospects in M.Obertino s talk
Other R&Ds on timing sensors } Micro channel plates: } not immune to B-field, but might be viable for LHCb } ~2-3 ps as secondary emission and amplification device } ~7% efficiency to MIPs, full efficiency to (pre)showers Glass+ALD 5x5 cm 2 16 Efficiency 1.9.8.7.6.5.4.3.2.1 imcp double-layer MgO imcp triple-layer imcp triple-layer imcp double-layer SEE imcp double-layer 1 1.2 1.4 1.6 1.8 2 Normalized voltage (V/V [ A.Ronzhin et al, Nucl. Instrum. Meth. A795 (215) 52 57 ] Efficiency [ L.Brianza et al. Nucl. Instrum. Meth. A797 (215) 216 221 ] th) 1.8.6.4.2 imcp double-layer imcp triple-layer imcp triple-layer imcp double-layer high SE 1 2 3 4 Absrber thickness (X )
Summary } Timing resolutions that are a factor 5-1 smaller than the timing spread of the beam spot open up new capabilities } Possible with both light and charge collection technologies } Dimensions matter a lot! Requires the technological capability to cover square meters of surface with small, thin tiles (1-1 mm 2 ) } Options for calorimeters: Correct vertex association requires vertex-t information I. Shower Max dedicated layer(s) embedded in the EM calorimeter or from the full longitudinal EM energy profile Additional Timing Layer a low-mass accompaniment to a silicon tracking system situated in front of a calorimeter system II. Pre-shower front compartment of the electromagnetic calorimeter - balancing low occupancy MIP identification with EM showering } Specific choices informed by radiation levels, integration and schedule constraints - make your choice and optimization! 17
Appendix 18
Luminous region 19
Relative Fake Rate Examples (w/o timing information): the anti-luminosity effect } CMS Upgrade Scope document: } [CERN-LHCC-215-19, LHCC-G-165] } VBF H!ττ requires >4% more luminosity at 2 than 14 PU } Jet fake rate and E miss T resolution } Searches with E miss T less sensitive at 2 PU than 14 PU 1.5 1.4 1.3 1.2 1.1 1.9 CMS Simulation Phase II, PU 2 Phase I, PU 5.8 1 2 3 4 5 2 Events/5 GeV 5 1 4 1 3 1 2 1 1 1-1 1 CMS Phase-II Simulation PUPPI, Phase-II w/o extended tracker 14PU PUPPI, Phase-II 14PU PUPPI, Phase-II w/o extended tracker 2PU 14 TeV E miss T spectrum PUPPI, Phase-II 2PU 38 5 Comparative Pe 14 TeV Jet background 2 PU 14 PU Jet η (GeV) lar component of hadronic recoil to Z boson, 1 χ m 2 4 6 8 1 12 E T (GeV) 2 1 14 TeV, PU = 14/2 CMS Phase II Delphes Simulation 8 6 4 2 5σ Discovery Reach χ ± χ 1-1 3 fb Phase II, 14PU -1 3 fb Phase II, 2PU -1 3 fb Phase II, 14PU, No Tracker Extension 2 W χ 1 H χ SUSY with E T miss 1 2 4 6 8 1 m χ ± 1 = m χ 2 (GeV) Figure 25: Effect of detector pileup on the sensitivity for the SUSY search fo miss 2 PU 14 PU 14 PU 2 PU
HGCal front end Signalaftershaper ADC ToT Time over Threshold (ToT) Buffer waiting L1 trigger accept 12.5 slatency 512eventsN32 bits=16.4 kb /channel Powerconsum ption:2mc /channel 21
UFSDs - Sensors } Above gain ~2 resolution is determined by charge non-uniformity (Landau fluctuations) 22