1. SLHC Physics and B tagging Joe Incandela University of California, Santa Barbara 10/12/06 Joe Incandela University of California, Santa Barbara 10/12/06

2. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 2 Outline The SLHC Physics Argument (cf. Eur. Phys. J. C39 (2005) 293) The physics case as from the viewpoint of the tracker Tracking and tagging at high luminosity ATLAS B tag study CMS Heavy Ions: Lepton track triggers B tagging in CMS @ LHC. B tagging in CMS @ SLHC: Does it become less important? What would be needed to make it work as well as at LHC? Some remarks

3. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 3 Disclaimer This talk was put together on short notice and mostly in an airplane… Not intended as a final word on anything… In the spirit of a workshop – it is meant mostly as a point of departure for more work and discussion

4. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 4 SLHC Physics P.Allport @Hiroshima ‘06

5. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 5 As viewed from the tracker Can divide SLHC physics into several relevant categories 1. Relies on tracking and/or b tagging 2. Requires absence of a track and/or b tag 3. Requires both 1 and 2 E.g. Vector boson resonance Need to identify high energy leptons Veto events with b tags to help eliminate tt background If no light higgs seen, this will be a major emphasis of SLHC

6. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 6 Leptons are top priority Multi-boson couplings & Higgs pair production Both rely upon efficient detection of leptons Leptons should be the 1 st priority of SLHC tracking All new ground. up to 6 l final states! Only sensitive if H above threshold to decay to VB pairs- 170 ≤ M H ≤ 200

7. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 7 The case for track triggers @ L1 From Gianotti et al. Eur. Phys. J. C 39, 293-333(2005) Larger event size due to higher occupancy means that 100 kHz L1 rate will probably need to be maintained…”such a strategy… implies raising transverse momentum thresholds on candidate electrons, photons, muons, etc. and using less inclusive triggers…”

8. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 8 Higgs Discovered Higgs at LHC At SLHC interested in rare decays, couplings to fermions and bosons. H  Z , H   +  - … H  and WH  l Comparison tells us about Htt coupling since gg  H proceeds via top loop Or if SUSY seen, we would want to extend our reach for a 2 nd and heavier SUSY higgs Need b/tau tagging.

9. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 9 Degradation of b tagging & Electron id No studies by CMS yet. ATLAS has studied current detector w/full simulations Mistagging rises x3 to x8 for fixed 50% b tag rate Rate of jets faking electrons at fixed 80% electron efficiency nearly doubles.

10. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 10 B tagging in CMS Very similar to Tevatron tagging Track tags Jet probability Number of tracks (2 or 3) above some impact parameter significance Vertex tagging Effectively similar in many ways to CDF SECVTX algorithm but developed in a much more intelligent way* * ( I can say this since I was one of the original developers of the CDF algorithm)

11. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 11 Inputs to Combined B tag Algorithm Jet Reconstruction: Iterative Cone Algorithm with cone size = 0.5 Track Reconstruction: CombinatorialTrackFinder ≥ 8 hits in total (pixel + strips) and ≥ 2 pixel hits pt > 1 GeV/c χ2/dof < 10 d xy < 2 mm (transverse impact parameter) Vertex Reconstruction: Primary Vertex: Global Reconstruction PVFPrimaryVertexFinder with reduced pt=0.7 GeV/c for tracks See CMS AN 2005/043 (C.Piasecki, C.Weiser et al.) Cone based association of tracks to jets: ΔR < 0.3

12. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 12 Two Definitions 1) “Physics” Definition: Match reconstructed jets to “initial” partons from the primary physics process (within ΔR < 0.3 of reconstructed jet cone) e.g. For tt the initial partons are: 2 b jets from top decays 2 non-b jets per hadronic W decay & no initial gluon jets jets from radiation are not matched with full efficiency Gluon jets splitting to c- or b- quarks are labeled “gluon” 2) “Algorithmic” Definition: Try to find the parton that most likely determines the properties of the jet and assign that flavour as true flavour here, the “final state” partons (after showering, radiation) are analysed (also within ΔR < 0.3 of reconstructed jet cone) jets from radiation are matched with full efficiency if there is a b/c within the jet cone: label it as b/c otherwise: assign flavour of the hardest parton

13. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 13 The Algorithm Inclusive vertex reconstruction in jets using the “Trimmed Kalman Vertex Finder”; select secondary vertex candidates: 100 μm < 2.5 cm Significance (L xy /σ) > 3 Invariant Mass of tracks in vertex candidate < 6.5 GeV Reject if vertex has two oppositely charged particles with invariant mass within 50 MeV of K 0 mass 3 Categories: depend on result of inclusive vertex reconstruction: 1. “RecoVertex”: at least one accepted SV candidate found 2. “PseudoVertex” : built from tracks incompatible with the primary vertex (d/  > 2), if at least two such tracks are present 3. “NoVertex”: the rest

14. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 14 Vertex Categories b c uds QCD 50-80 |η| < 2.4

15. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 15 Input Variables I Enter for all categories into the final discriminator Furthermore: -sort tracks in decreasing order of IP significance -compute mass for tracks -look at IP significance of track pushing the mass above threshold related to charm hadron mass (here: 1.5 GeV) “RecoVertex” “PseudoVertex” “NoVertex” b c uds QCD 50-80 |η| < 2.4 Lifetime signed 2D Track Impact Parameter Significances

16. C.Weiser A Combined SV Based B-Tagging Algorithm in CMS CMS Phys. Meeting 6/12/2005 P 16 Input Variables II Additional secondary vertex related variables for category 1 (“RecoVertex”) b c uds inv. mass of charged particles at SV multiplicity of charged particles at SV PV-SV σ PV-SV QCD 50-80 |η| < 2.4

17. C.Weiser A Combined SV Based B-Tagging Algorithm in CMS CMS Phys. Meeting 6/12/2005 P 17 Input Variables III Additional secondary vertex related variables for category 1 (“RecoVertex”) Enters for n tracks b c uds fractional charged energy at SV IP sign. of first track above charm mass rapidities of charged particles at SV QCD 50-80 |η| < 2.4

18. C.Weiser A Combined SV Based B-Tagging Algorithm in CMS CMS Phys. Meeting 6/12/2005 P 18 Input Variables IV Additional secondary vertex related variables for category 2 (“PseudoVertex”) b c uds inv. mass of charged particles at SV multiplicity of charged particles at SV fractional charged energy at SV rapidities of charged particles at SV IP sign. of first track above charm mass QCD 50-80 |η| < 2.4

20. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 20 b c udsg QCD 50-80 | η | < 2.4 b c udsg Plots have been obtained by scanning the cut on the discriminator Final Discriminator II

21. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 21 QCD 50-80 GeV

22. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 22 Misidentification efficiencies for fixed b-tagging efficiency of 50% | η | < 2.4 QCD 50-80 pt non b-jet efficiency loss of tracks in this bin! ● uds * g ▲ c η Dependence On PT and η

23. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 23 Heavy Ions as a proxy for pp@SLHC

24. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 24 HI Algorithm Default pp algorithm with following modifications: 1) Trajectory Seed Generation Three pixel hit combinations (Pixel triplets) Primary vertex constraint 2) Trajectory Building Includes all material effects multiple scattering energy loss Special error assignment to merged hits 3) Trajectory cleaning Allow only one track per trajectory seed: best  2 4) Trajectory Smoothing Final fit with split stereo layers

25. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 25 Acceptance Require  8 strip layers (~12 hits) and 3 pixel layers. Geometrical acceptance ~80%

26. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 26 Track quality Cuts More than 12 hits on track (stereo layer => 2 hits) Require fit probability > 0.01 (Cut on compatibility with primary vertex) Good reconstruction

27. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 27 Fakes Fakes substantially higher than at LHC, as seen by ATLAS. nhit > 12 pchi2 > 0.01 Efficiency o Fake Rate Efficiency o Fake Rate looser

28. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 28 Trajectory Building Number of candidates drops fast as you move to larger radius even though the occupancy does not fall as quickly. Track gets more refined and so road narrows…

29. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 29 High efficiency setting Algorithmic Efficiency and Fake Rate vs 

30. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 30 Algorithmic Efficiency and Fake Rate vs  Low Fake Rate Setting

31. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 31 CMS Subtraction Would all strips continue to need to be read out?

32. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 32 Go back to that occupancy plot… Mismatch at system boundaries Layer 4 appears pretty useless here. 3 cm strip would cut occupancy to under 10% Pixel layer would likely ice the cake Layers 5-13 Shorter strips but fewer layers to compensate for material? A factor of 3 reduction in strip length would do a lot for this plot.

33. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 33 ATLAS Granularity Guide Using mainly strips, they’d get substantially better granularity than CMS has now

34. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 34 But granularity can be wasted… Fine Granularity is necessary but not sufficient… It is wasted if there is substantial multiple scattering, radiation, and secondary particles generated in material interactions. Material can set an effective granularity if one is not careful Material reduction throughout the tracker would enhance our effective granularity now, without changing anything else.

35. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 35 Multiple Scattering Now Cucciarelli et al. CMS Note 2006/026

36. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 36 Well this is a workshop… I don’t have answers… want to 1 st frame the questions correctly Main points: Leptons 1 st displaced tracks 2 nd on priority list (just in case push comes to shove…) Granularity can be improved substantially without necessarily using pixels - but it will be meaningless if the material budget is not reduced. Some thoughts Some technologies could run warm (e.g. n-in-p discussed in Mara Bruzzi’s talk tuesday) … This might allow us to eliminate much of the cooling related mass in the tracking volume. Super thin is in. We need to think more like e+e- people regarding our material budgets Recognizing that material can alter the effective granularity of a tracker, we should carefully consider the possible benefits of a reduction in the number of layers ! We need to do some serious studies of these issues Triggering was not covered in this talk, but is important, particularly for leptons

37. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 37 Tangent-Point Reconstruction α J.Jones Imperial College London (Perugia Workshop) 37

38. More stuff

39. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 39 construction costs Upgrade cost  ~ 200MCHF Broad brush estimate reported to CMS MB/CB & DG ~60-70% tracker related plus staff costs (significant) Inner Tracker25-30 MCHF Outer Tracker90 MCHF Level 1 Trigger15 MCHF DAQ10 MCHF Other Front Ends5-10 MCHF extra costs 10ns/15ns20-30 MCHF Infrastructure15 MCHF G. Hall, Imperial College CMS Tracker Meetings, CERN - August 2006

40. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 40 Roadmap from previous workshop Installation of modest system at t = t o + 5y may be possible Lower cost and risk Allows trial of components or devices, which may still evolve May be possible to react to LHC conditions machine, experiment or even discoveries An evolutionary approach to replacing full tracker? Ideas are still to gel but must do so soon CMS proposes common EoI (2006), and LoI (mid-2007) R&D proposals to be evaluated by CMS, and approved/encouraged G. Hall, Imperial College

41. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 41 You can’t always get what you want… But how much increase in granularity is actually needed? Present microstrip occupancies are 0.5% - 2.8% in barrel CMS NOTE 2002/047 G. Hall, Imperial College

42. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 42 Potential synergy with ATLAS ATLAS & CMS have exchanged speakers at several workshops Initial phases of LHC R&D were common Although many similarities, also important points of divergence eg sensor design, electrical & optical interfaces, analogue/digital, DAQ design,.. Possible common efforts Sharing ASIC processing runs (in CMS & CERN done well for 0.25µm CMOS) Advantageous to share circuits, evaluate technology and adopt common standards Share development of common SLHC systems Optical links and Timing-Trigger-Control system are prominent items Common effort on power provision - eg DC-DC conversion? Dialogue with machine Agree clock speed, verify current systems, information about machine operation Special tooling removal and installation of irradiated systems in irradiated environment Information exchange via regular meetings Annual LECC workshops are one common forum for electronic R&D Comparison of cooling system performance might be profitable G. Hall, Imperial College

43. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 43 Cooling costs Using power has heavy material cost For present pixel system Power in ~4% Power out ~29% For microstrips Cables ≈ Cooling Cables + Cooling + Support ≈ 2x (Sensors+ Electronics)

44. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 44 Noise & granularity Leakage current shot noise also determines element size Noise scales ~ (area, time, fluence, shaping time,...)1/2 How much can be gained by cooling? (ATLAS discuss -35°C) I leak will be more significant power burden so must be contained thermal runaway is increased danger Is required lifetime again 10 years?

45. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 45 Seeds

46. SLHC Physics and B-tagging : FNAL Pixel Workshop, October 12, 2006 J. Incandela (UCSB) 46 Performance of the Track Reconstruction Match Reconstructed tracks to MC input on a hit by hit basis. (Event sample: dn/dy ~3000 + one 100GeV Jet/Event) Momentum Resolution Transverse Impact Parameter Resolution Longitudinal Impact Parameter Resolution