High Precision Time-of-Flight for the CMS Phase II Upgrade

1. High Precision Time-of-Flight for the CMS Phase II Upgrade Adi Bornheim, Artur Apresyan Caltech Timing Simulation

2. The Goal • Use Time-of-Flight information for the CMS Phase II detector to mitigate the effects of Pile-Up. • Benchmark applications : • Use TOF of jet constituents to suppress activity from PU vertices. Example : VBF jets. • Use TOF to check compatibility of photon vertices. Example : Photons from Higgs decays. • Use TOF to check compatibility of individual rechits coming from a common primary particle. Example : Timing based ECAL clustering cleaning. Timing Simulation

3. The Challenge In this chapter we describe the physics environment (kinematics, multiplicity) in which we need to perform the high precision timing measurements. • We need to measure the arrival time of particles from pp collisions at 14 TeV with 140 simultanious pp collisions spread out over an interaction region of about 10 cm (FWHM). • 14 TeV proton collisions : H decay photons and electrons have typical energies from around 10 GeV (low pT electrons from HZZ to around 60 GeV photons from H in the barrel to >100 GeV for photons and electrons in forward direction. • Energy flow :  plot • Multiplicity :  plot • Vertex timing spread : 200 ps Timing Simulation

4. Energy flow (FWD-10-011) Timing Simulation

5. Track multiplicity (FWD-10-008) Timing Simulation

6. Particle dN/dη, dN/dpT at 0.9, 2.36 and 7 TeV • arXiv:1002.062, JHEP 1002:041,2010 (0.9 and 2.36 TeV) • arXiv:1005.3299, submitted to PRL (7 TeV) Timing Simulation

7. Benchmark plots for HL_LHC Possible detector level benchmark plots for high pile-up : ECAL endcap PU energy in clusters from Zee and We. Jet energy PU corrections. Example EE electrons : 2% PU energy pick-up for 20 PU.  14% for 140 PU ! Example JET PU corrections : 40% PU energy pick-up @20 PU @ 50 GeV  280% for 140 PU !

8. Implications • Need to have sufficient granularity to allow timing measurements of jet constituents. • Need detailed dn/deta info as a function of pT. • Need to be able to associate a energy/momentum measurement to the timing measurement. • Need to measure timing for high energy particles (benchmark applications 2) and 3). • The characteristic length scale for interactions of EM type particles with matter is X0 which is : • 7⁄9 of the mean free path for pair production by a high-energy photon • a high-energy electron loses all but 1⁄e of its energy by bremsstrahlung • A TOF detector for photons needs to have sufficient thickness to have a high detection efficiency. Timing Simulation

9. Timing Simulation

10. R&D Program • Toy MC simulation (Sepehr studies) : • Study geometric boundary conditions. That is, how big is the actual TOF difference assuming a certain timing detector geometry, vertex distribution and vertex timing distribution, PU conditions, etc. . • Geant level studies (Yong studies) : • Study the impact of the interactions of the particle on its TOF. Timing spread due to showering process before and in the timing detector. • Study on MC true level the actual overlapp of particles from PU and the hard interactions (spatial, time, momentum). • Hardware R&D (Cosmic Timing Test Stand) : • Test and optimize precision tming setup. • Hardware R&D (Combined Test Beam) : • Use exisiting hardware to validate Geant level studies and measure what is not reliably simulated (eg. light propagation times in complex geometries such as shashlik setups). • Explore actual limits of thin and thick timing detectors. Timing Simulation

11. Cosmic timing test stand (CTTS) 1 8 4 7 7 5 6 4 2 3 1) : Cosmic muon (solid line), cone of acceptance adjustable via trigger counter geometry 2) : `Thin` high precision timing devices (eg. Henry plates), minimum one, idealy three 3) : Readout for 2), triggered by 5), interfaced with 8) 4) : Szintillator trigger counters for cosmic muons 5) : Trigger logic for 4) 6) : LYSO crystal 7) : twosided readout, eg. SiPMs 8) : Readout for 7), allows differential timing measurements of two sides. Timing Simulation

12. CTTS program • A) Stage one : • 4) : Szintillator trigger counters for cosmic muons • 5) : Trigger logic for 4) • 6) : LYSO crystal • 7) : twosidedreadout, eg. SiPMs • 8) : Readout for 7), allowsdifferential timing measurements of twosides. • Milestones : • Put together setup and time in cosmic muon trigger. • Optimize timing precision of 7) (noise reduction, different photo sensors, readout) • Measurecorrelationbetween muon crossing location along the crystal and time difference determine timing resolution of cosmic muon setup. • B) Stage two : • 1) : Cosmic muon (solid line), cone of acceptanceadjustable via trigger countergeometry • 2) : `Thin` high precision timing devices (eg. Henry plates), minimum one, idealythree • 3) : Readout for 2), triggered by 5), interfaced with 8) • Milestones : • Synchronize with muon crystal setup above and check muon signal in 1) • Benchmark timing of muon crystal setup against 1) Timing Simulation

13. CTTS Simulations and Estimates Timing Simulation

14. Timing Simulation

15. Timing Simulation

16. Geant4 simulation A. Apresyan

17. Setup • Trying to estimate how effective will the setup with Henry device be • Simulate muon passage through the glass window • Estimate total photon yield and energy spectrum from Cerenkov • Initially started with the program called G4BeamLine • Has a nicer interface to Geant and simpler • But turned out does not handle optical photons • Switched to full Geant4 Schematic of the fast timing detector. A relativistic charged particle produces Cherenkov light in the window. This radiation is converted into electrons by a photocathode. The electrons produce a shower in the micro-channel plates (MCP), and the electrons are deposited on the segmented anode to be detected.

18. Setup • Implemented the detector volume as: • Borosilicate glass of 3mm thickness • Refractive index of 1.538 and Absorption length 4.1-5.5cm (as a function of photon energy) • Glass is immersed in Air • Surface is also implemented to have the reflections and scattering • Shoot a beam of muons at 1GeV at the front face • See how many optical photons are produced • Optical photon: 2.47-3.43 eV

19. Event displays • To validate the setup, increased the thickness to 3cm • Beam comes from left • Clearly see the Cerenkov cone • Seems to be a lot of backscatter at the back face, facing the cathode?

20. Number of optical photons • This shows the total number of photons produced per event • Here the thickness is set to 3mm

21. Simulation adding a PMT layer on the back of the glass

22. Next steps • Now adding the rest of information into ntuples • Production mode, number of escaping photons, energy spectrum • Validate against ionization energy losses from MIP in ECAL • Move to more complicated setup, add a crystal in front of the Glass

23. Here the thickness is 3mm

24. Timing Simulation

25. Timing R&D Setup 6 3 1 2 7 4 5 1)-3) : `Thin` high precision timing devices (eg. Henry plates), minimum one, idealy three 4) : Shashlik prototype with dual side fiber readout (once available and test with 5) are done) 5) : Solid crystal with dual side optical readout (first test with this, since simpler than 4) ) 6) : Illustration of shower depth profile 7) : Vertexing device to have a precise impact point and angle (optional for first test) Timing Simulation

26. Specific tests • Measure the timing precision of the Shashlik or solid crystal calorimeter, using the thin precision timing device as a reference. • Important `fundamental` limitations of the calorimeter : • Shower fluctuations. • Light propagation time. • Use dual readout to measure the shower depth event by event. • Additional tests / modified hardware confiurations : • Insert timing device into Shashlik stack (only once Shashlik has been verified intially). • Insert/attach sensors to individual LYSO plates to measure shower timing profile. Timing Simulation

27. Timing Simulation

28. Summary Timing Simulation

29. Backup Timing Simulation

30. Motivation T : 10 ps • Motivation : Study if a TOF detector with timing resolution order ps can improve photon and jet vertexing and PU mitigation. • Example with simplified toy MC : Jet vertex resolution (di-jet events, only rechits with E>4 GeV, PU rechits from MinBias data, collision time simulation). • Order mm vertex resolution in forward direction without PU. • Significant resolution degradation without PU filtering. • Next step : Rechit filtering based on timing. Timing Simulation

31. Proposal for CMSSW • Exploit code infrastructure of ECAL time reco (time per rechit). For FullSim : • Replace ECAL rechit timing in the MC with the true time of a geant particle entering the crystal front face. Smear true time as desired (1 ps, 10 ps, 100 ps). • Need some thinking in case of multiple hits (energy threshold ?), association of geant particle with ECAL energy and some other technicalities. • Started some discussion with experts, have one person available for a few weeks to work on an initial implementation. Additional expert advise more then welcome . Question : • To what extend is this scheme transferable to FastSim ? Timing Simulation

32. Current Work (Yong Yang) Two possible strategies : • Define TOF geometrically by arrival of a given particle in the timing detector cell. • Define TOF by first geant hit in the timing detector. We currently pursue the second option. This implies that we study already some aspects of the detector itself. It seems technically easier to us right now. Pursueing the first option requires a proper association of the arrival time with the interaction in a given volume. Both require some more thinking. Timing Simulation

33. Geant4 Timing Studies in CMSSW Timing Simulation

34. Time scales • We have a first algorithm to get a ^true^ arrival time. • Expect a few more weeks to arrive at a more solid solution. • We need some framework help to transport the relevant info to the ECAL rechit level. Timing Simulation