CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 1 LHCb : what to do with the first month of data taking

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 2 Premessa LHCb plans to set up a detailed strategy for the detector and trigger startup over the coming years. My report summarizes the current ideas being discussed with some detector and trigger experts and must be taken as a starting point

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 3 Outline detector and trigger general structure subsystems and calibration requirements trigger description and startup a first look at physics

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 4 LHCb detector  p p ~ 200 mrad ~ 300 mrad (horizontal) 10 mrad Inner acceptance 10 mrad from conical beryllium beam pipe

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 5 LHCb vertex region  VErtex LOcator around the interaction region, vertices reconstruction

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 6 LHCb tracking  Tracking system and dipole magnet: to measure angles and momenta (together with VeLo)

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 7 LHCb P-ID  Two RICH detectors for charged hadron identification

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 8 LHCb calorimeters  Calorimeter system to identify electrons,hadrons and neutrals and used in the L0 trigger e h

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 9 LHCb muon detection  Muon system to identify muons and used in L0 trigger 

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 10 Trigger strategy 40 MHz HLT: Final state reconstruction Calorimeter Muon system Pile-up system VErtex LOcator Trigger Tracker Level 0 objects (Muon) Full detector information Level-0: p T of , e, h,  Level-1: Impact parameter Rough p T:  (p)/p~ 20% 200Hz 1MHz 40kHz Level-0 : Use large B mass signature hardware system with fixed latency (4  s) Level-1 : use B mass and lifetime signature Software analysis on reduced data from only few detectors. Run on a PC farm in common with HLT (~ 1800 CPU total) High Level Trigger (HLT): Software.The complete event is reconstructed with almost final accuracy Selection of interesting physics decays

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 11 VertexLocator -Silicon detector -21 stations with alternated R-  segmentation - sensitive area starts at 8mm from the beam -retractable to 3cm during refill (the two halves open along the horizontal axis) -used in L1 trigger: initiate tracking to compute IP and P t 2 Pile-Up layers (only r sensors) to veto multiple interactions at L0 Interaction point  z =5.3cm ~ 1m ~10cm

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 12 Top view of the VeLo system

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 13 VeLo structure: R-  sensors Pitch  m Stereo angle 10 ° -20 ° 45° sector Pitch  m  R: y x y x angular coverage of each half disc = 182 o total # channel ~ analogue readout

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 14 VeLo alignment figures - Silicon strips are positioned on disc halves with <5  m - relative position of disc halves, within a half VeLo, surveyed before installation at  x),  y  ~ 5  m,  z)~ 20  m precision in air and at room temperature. to be checked on beam with tracks since vacuum and cooling may affect it Goal is an hardware alignment of  x),  y  ~ 20  m for L1 trigger to work - since VeLo retracts at each machine refill the two parts are re-positioned with a precision of <~100  m, this can be checked with a few seconds of data taking thanks to small overlapping surfaces - final transverse software alignment precision for event reconstruction ~ 5  m

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 15 VeLo test beam activity before installation A complete Half VeLo module should be tested on 120 Gev beam -measure, and physically correct, misalignment among half discs -full test of vertex reconstruction capability (target placed in front of the velo to simulate interactions),including software - this is very important for its own survival: at LHC startup beam position can be estimated by VeLo itself reconstructing primary vertices at its retracted position before getting close to it (VeLo can also determine the beam position before collisions with beam-gas interactions)

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 16 Tracking system 1 station upstream the magnet and 3 downstream dipolar magnet: Bl=4Tm(*) each station is composed of 4 layers : x,+5 o,-5 o,x TT: silicon technology T1-T3 mixed technology (according to occupancy) Inner Tracker: silicon strips Outer Tracker: straw tubes (*) installed&switched on successfully

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 17 Tracking system: TT Station four layers: distance between 2 layers 30cm total silicon area 4x2m  m pitch Build with sensors each 9.6x9.4cm 2 bonded to form strips 10cm–40cm long 143,000 FE channels used in L1

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 18 Tracking system: T1-T3 IT silicon sensor: area 11x8cm 2 strip pitch 198  m Max occupancy 1,5% 130,000 channels OT 45 0 Two straw tubes (5mm Ø) layers staggered by 5.25mm for no dead space 56,000 channels Resolution ~ 200  m with drift-time measurement Occupancy max < 7%

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 19 Trackers alignment Silicon trackers strips positioned with ~5  m positioning precision of detector boxes at survey~1mm relative precision needed at L1 for VeLo-TT ~<200  m, to be done with first tracks final software alignment aims at about 10  m for complete event reconstruction Straw tubes installation precision at survey ~1mm working without drift time information at the beginning (  m),more ghosts but pattern recognition with reasonable efficiency possible.Found tracks used for alignment and r-t calibration

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 20 RICH system 2 Rich with 3 Cherenkov radiators Need of PID in large p range GeV P threshold (GeV)AerogelC 4 F 10 CF 4  K RICH1 RICH2

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 21 RICH mirrors alignment(1) RICH system is used in the offline analysis (its use in HLT is under development) Rich1 has 4 spherical+ 16 flat mirror segments Rich2 has 58 spherical and 40 flat mirror segments Mirror tilts can affect the resolution on the Cherenkov angle Software corrections are possible with a precision of 0.1 mrad provided a mechanical precision of 1 mrad is obtained during mounting (attainable with laser alignment). This limits the fraction of photons which cannot be attributed unambiguously to one mirror (for which no correction possible) avoiding the deterioration of PID performance. Alignment is much more critical for Rich2

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 22 RICH mirrors alignment(2) Need for a working tracking, ~1k well reconstructed tracks/mirror needed,<~100Kevents

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 23 ScintillatingPadDetector& PreShower : two 15 mm scintillating pad layers + 2.5X 0 lead in between Electromagnetic calorimeter: “shashlik” 2mm lead + 4mm scintillator, 25X o  (E)/E=10%/√E  1.5% Hadronic calorimeter.: iron-scintillator plates, 5.6   (E)/E=80%/√E  10% Calorimeter system(1)

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 24 Calorimeter system(2) 5952 ch 4 cm 12 cm 6 cm 26 cm 13 cm 1468 ch SPD : multiplicity, e/  separation PreShower: e/h separation ECAL: e,  E,E t ) measurement } HCAL: hadrons (E,Et)

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 25 Calorimeter calibration Ecal modules pre-equalized with cosmics Hcal modules pre-equalized with 137 Ce a sub-sample is energy-calibrated on test- beam, calibration transported on the whole detector with momentum well reconstructed tracks PM gain stability monitored with pulsed LED ( checked with photodiodes). This serves also as signal time synchronization

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 26 Muon system(1) 5 Stations + 4 iron filters 1380 chambers: 1368 MWPC + 12 GEM => 435m 2 trigger imposes high efficiency/station since 5 stations/5 are required (>99%)  4 (2 in M1) gas gap ORed (each gap has  >~95%),

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 27 Muon system (2) Large (10x8x7 m 3 ) system 120K FE channels ORed to form 26k logical channels sent to L0/DAQ Up to max 48 physical channels make up a logical channel Granularity: Min 6.3x31.3 mm 2 (M2R1) Max 250x310 mm 2 (M5R4) One quadrant of Station 2 A high precision space alignment is not needed for the muon system but…… 

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 28 Muon system time alignment(1).... an accurate time alignment is mandatory to ensure high L0 muon trigger efficiency MWPC time spectrum of one physical channel (OR of 2 gaps) with cosmics. 25 ns If not well centered significant losses are possible

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 29 Muon system time alignment(2) Muon electronics is designed to provide synchronization tools: ASD chip (Carioca) DIALOG(*) chip FE board SYNC chip ODE board The SYNC chip accumulates hits from the FE -in BX bins (25ns) -inside the BX with a 1.5ns resolution TDC With SYNC histogramming facilities the total signal shift can be determined as M*25ns+K*1.5ns - M is corrected on the SYNC itself while K is corrected with an adjustable delay present on the DIALOG chip (*) DIALOG performs the logic OR to form logical channels

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 30 Muon system time alignment(3) Coarse alignment: hits in 16BX around the beginning of a full bunch crossing sequence whose timing is given by the accelerator (16 entries/LHC orbit) 16 cycles BX identified Residual misalignment BX number Before FTO correction BCn+2BCn+3BCn+2BCn+3 Phase measurement (25 ns gate) On-chip histogram BC cycle FTO determination and correction via front-end Programmable Delays (DIALOG chip) FT alignment Statistics in ~1h Statistics in <~1h Fine alignment

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 31 Muon system time alignment(4) Peak position for 16 FE ch forming 1 logical channel 5 ns 25 ns It might be not true for logical channels formed by signals coming from different chambers no muon trigger needed (all hits are usable) no DAQ needed, histogram analysis through ECS alignment possible at the physical channel level (masking facility in DIALOG chip). In principle, in fact, each FE channel must be independently centered in 25ns. Cosmics test shows, however, that FE channels inside a logical channel are in time within < 5 ns

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 32 Muon system time alignment(5) After L0 has started (even with reduced efficiency) refinement can be done with offline analysis requiring channels with a muon hit (TDC information in DAQ) All hits Muon hits MC

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 33 L0Trigger  >90%  K decay Nominal threshold HCAL clusters Select local maxima in ECAL and HCAL towers Use SPD and PS to separate e/  /h MuonCalo

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 34 L0 Trigger(2) HCal dominates Muon dominates ECal dominates ZBZB ZAZA RBRB RARA Z PV 2 silicon R-stations Possible cut: retain >98% of single and reject ~60% of multiple Gain of 30-40% of single bb- events at optimal luminosity PileUp veto (to reject multiple interactions) L0 result.

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 35 L1 Trigger - track reconstruction in VELO Velo 2D reconstruction Primary vertex search VELO sensor design optimized for fast track finding: B’s move predominantly along beam-line  impact parameter in RZ-view 1.Straight line search in R-Z view, forward and backward tracks ~ 58 (+ ~30 backward) tracks 1.Vertexing, σ Z ~ 60 μm, σ X,Y ~ 25 μm 2.Select tracks with high impact parameter, 0.15 to 3 mm about 8.5 per event 3.Full space tracking for those tracks 4.Extrapolate 3-D VELO-tracks to TT, use fringe field for P estimation dp/p=20-40%,  (B-tracks)=94%

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 36 L1 Trigger–match large IP with L0 candidates Match with high Pt L0 calo and muon tracks Study ongoing to include muon information Signal/background separation by Pt and impact parameter

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 37 HLT Trigger Confirm Level-1 decision –Complete Velo 3D tracking –Primary vertex search –Select from the 3D Velo the large IP tracks, measure accurately the momentum, confirm the decision (same variable) Extend Velo tracks across the magnet  (p)/p ~ 0.6 % Gain a factor 2-3 without significant loss Fraction of CPU time budget Input 95% confirmed Physics selection (including lepton–ID but not Rich,yet) It has been checked for a few representative channels that a loose exclusive selection brings the rates down to Hz per channel

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 38 Trigger startup(1) Assumptions: the subsystems are commissioned prior to data taking with the LHC beam ( noise, missing channels etc) one circulating beam cannot be easily used since earlier beam is most likely in the wrong direction but it can be helpful for VeLo something can be done with beam-gas interactions when the two beams circulate (alignment of the tracking elements). we have assumed we will use the first collisions 5 2 N coll /BX R(MHz) LHC scenario: L~10 32 cm -2 s ns bunch spacing E beam =6 TeV ~ 5-6 MHz with >=1 collisions

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 39 The transfer line with the right direction (TI2) will be installed in 2007

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 40 Trigger startup(2) Muon system time is aligned as described VeLo system is retracted Calo system is operational Set-up an interaction trigger with minimum energy released in HCAL and/or multiplicity in SPD. (SPD can be used also to select particularly “clean” events) Take minimum bias data O(10 6 ) and check of: - m.b. rate - measure beam position with VeLo retracted - check correct functioning of L0 triggers - measure Pt distributions at L0, rates.vs.Pt cut - when VeLo closed, pile-up veto efficiency can be checked - take data also with magnet off to perform relative of tracking elements : aim at an alignment sufficient to let L1 work

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 41 Trigger startup(3) L1 switched on check IP distributions check capability to match high Pt tracks check L1 and HLT rates A large fraction of the data processing during this startup phase could be done by the L1/HLT CPU farm at the pit, assuming the ~full CPU power is available at t 0 Finally, as soon as we have a reasonable trigger, we can start taking useful data and first signal to look for is obviously J/   

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 42 Look for J/    J/  is selected with a dedicated dimuon trigger 2 VeLo tracks coming from a common vertex and connected to 2 L0muon candidates (Pt1+Pt2)>1.3GeV no IP cuts, M inv window around J/  mass We can have a ~70Hz of J/  signal after L0&L1&HLT(*) with B/S~2-3 (*) J/  HLT

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 43 J/    If we are lucky a J/    sample of ~3x10 7 could be gathered in 2 weeks running with a J/    trigger With it we can do a more precise tracks momentum calibration using J/  mass and vertex Once the detector is well aligned and calibrated we may start to look at physics:  (J/  prompt  mb  - a lot of b’s  J/  X Pythia gives: J/  b  J/  prompt  ~ 5% (CDF  probably effect of cuts  ~1.5Mevts b  J/  X

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 44 J/    Dimuon Trigger: Pt > 1.5 GeV per  ~2 million psi ~80% prompt ~20% B

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 45 b  J/    Statistics of some interesting B decay channels assuming nominal detector performance Channel J/  ident Full ev.rec. B 0  J/  K s 0  BR vis =1.9x Kevts5Kevts B 0  J/  K *0 (  K) BR vis =5.9x Kevts14Kevts B +  J/  K + BR vis =6.8x Kevts37Kevts B + c  J/  + BR vis <=6.8x10 -4 <=1Kevts<=500evt

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 46 B 0  J/  K *0 (  K) Control process, no CP violation B oscillation could be seen (with some tagging efficiency, maybe only  ) Stat.err. only, nominal tagging, no Bg (B/S=0.37)

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 47 B +  J/  (  )K + Tagging check Lifetime check (no IP bias in this sample) B + c  J/  + Very clean peak at 6.4 GeV Never observed

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 48 Conclusions A detailed startup procedure will be established in the coming years Trigger startup, sub-detectors calib. & align. (even if not final) feasible in the first month of data provided a lot of work is done before, also on test-beams Part of this work can be done without offline analysis, while the rest implies data taking and full reconstruction Some “not really physics results but showing our capability of doing physics can be achieved”

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 49 Backup

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 50 Trigger: possible improvement There is an on-going study to increase to ~ 2kHz the events rate to DAQ. 200Hz as baseline strategy + ~2kHz of events with a muon with high impact parameter On muon ”highway” triggered events b content ~ 50% Gain a sample of unbiased b (the companion of triggering b->  ) –useful for systematic study –Inclusion of b decay not easy to trigger on –Inclusion of b decay not of interest now but maybe in the future…

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 51 VELO Primary vertex resolution in bb events:  and || to the beams  z =44  m  x-y =8  m  IP =14  +35  /p T Impact parameter resolution vs 1/P T

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 52 Vertex Locator Primary vertex resolution in bb events:  and || to the beams  z =44  m  x-y =8  m  IP =14  +35  /p T Impact parameter resolution vs 1/P T

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 53 Track finding strategy VELO seeds Long track (forward) Long track (matched) T seeds Upstream track Downstream track T track VELO track Long tracks  highest quality for physics (good IP & p resolution) Downstream tracks  needed for efficient K S finding (good p resolution) Upstream tracks  lower p, worse p resolution, but useful for RICH1 pattern recognition

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 54 Tracking performance Ghost rate 0.5GeV Eff=94% P>10GeV  p/p = 0.35% – 0.55% Momentum resolution Ghost rate Track finding efficiency

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 55 Rich Particle Identification Bs->KK decay K)> = 88% K)> = 3%

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 56 K S reconstruction DD  =54% LL  =75% LU  =61% Ks decays inside Velo(25%) (LL+LU) between Velo TT(50%)(DD) behind TT (25%) Only first 2 cases reconstructed Low momentum track swept out by the magnet: (U type track) poor p resolution When both tracks are reconstructed the selection is reasonably efficient

CSN1, 16 Novembre 2004 Roberta Santacesaria, INFN Roma1 57 Flavour tagging sources for wrong tags: B d -B d mixing (opposite side) b → c → l (lepton tag) conversions… Tag  Tag (%) w (%)  eff (%) Muon Electron Kaon Vertex Charge Frag. kaon (B s ) Frag  (B)0.7 Combined B 0 (decay dependent: Combined B s trigger + select.) ~4.7 ~6 effective efficiency :  eff =  tag (1-2w tag ) 2