The LHCb Upgrade Chris Parkes Why Upgrade ? Trigger System Radiation Hard Vertex Detector Thanks to my LHCb Collaborators for advice/results/speculations… Vertex 2005, November 2005, Nikko VElo Superior Performance Apparatus

Dedicated B System CP violation Experiment Full spectrum of B hadrons: Bs system,All angles, sides of both CKM s Lots of events ! p 250 mrad 10 mrad Vertex Locator Dipole magnet Tracking system Calorimeters Muon system RICH detectors

R / Phi measuring sensors Velo Rôles Primary / b decay Vertex reconstruction Stand alone Tracking A principle tracking device for the experiment Second Level Trigger Fast tracking 1m R / Phi measuring sensors In vacuum Retract each fill Align each fill One set of half disks

Yet another B-Factory ? LHCb - Dedicated B physics experiment at LHC designed for precision study of CP violation and rare decays BTeV Cancelled BaBar Closure forseen Super-Belle ? Likely to be only B-Factory in LHC era LHCb  now 47 institutes in 16 countries > 600 authors But what is left to do after the B-factories?

What Did the 1st Generation B-Factories Do For Us ? Spectacular progress from the B-factories: Baseline measurement ACP (J/y KS) Agreement with Standard Model CKM Impressive range of additional measurements Significant constraints on  95% CL

Some Puzzles left by the B factories….. New physics in b s ? sin 2bb  s = sin 2bb  c? see in Bs  f f ? increased BR for Bs  m+m- ? Higher frequency Dms ? Larger CP violation in Bs  J/y f Bs–Bs oscillation Dms as Standard Model CDF or D0 measure ? If beyond SM… LHCb VELO / trigger required dms > 14.5 ps-1    xs > 21.1 (95% CL)

After 3 Years of LHCb….. 3 Years of LHCb data taking Bs mix 3 Years of LHCb data taking 1 day at LHCb = 100d at B Factory ! In event rate but hadronic environment…. Bs Oscillations measured SM <25 ps-1, CDF LHCb ms reach 68 ps-1  measured Theory error ~1% will be matched by LHCb ~ 5yrs  measured J/ K0 theory error < 1%, 1 yr statistical error sin(2) 0.02 LHCb: st = 43 fs

Limited by Statistics. B Physics after 2010 (What will NOT be known after 3 years of LHCb running ?) Precision Gamma (< 5 degrees) Theoretically very clean, error only ~ 0.1% ! etc. “Tree” only New Physics in D mixing? Improved vertex Resolution equiv. to more stats Limited by Statistics. Rare B-decays The other Triangle…. BsJ/ SM BR at most 3.5x10-9 ! 5 events per 2 fb-1 (S/N 3:1) Weak mixing phase  Proper-time resolution important 8 independent parameters to determine SM BR at most 1.5x10-10 !! Requires upgrade

But NOT Limited by LHC Pile-up at high luminosity Defocus LHC beams Limited by Statistics. Number of pp interactions/ beam crossing Pile-up at high luminosity B mesons identified by separation of primary interaction vertex and decay vertex (few mm) Displaced Vertex trigger 2nd level of triggering Multiple Interactions Limit Event reconstruction Defocus LHC beams ATLAS/CMS 1034 cm-2s-1 but LHCb 2x1032 cm-2s-1  most events have single interactions Could LHCb cope with higher Luminosity ? 1033 cm-2s-1 Extreme Radiation Environment in VELO

Increase Luminosity – Trigger System LHCb would have to cope with multiple interactions Existing 1st Level Trigger Veto on multiple interactions ! Trigger based on: High pT Muons & calorimeter clusters Trigger on BsJ/KK 5 X Yield Number of Triggered Events 4 X Yield 1st Level Trigger Rate (MHz) 3 X Yield 2 X Yield Std. Yield 4 of 10 benchmark LHCb channels have +- in final state Luminosity (x1032) Enhanced Rare B-decay Programme WITHOUT Trigger upgrade Radiation Damage, Occupancy

Hadron Channel Trigger Trigger on BsJ/KK 1.8 X Yield Hadron Channel Trigger High ET Trigger not sufficient Std. Yield 1st Level Trigger Rate (MHz) Number of Triggered Events 1st Level Displaced Track Trigger Latency 4s, 2s for data processing Luminosity (x1032) Massive use of FPGAs can allow us to make a Vertex trigger in ~2010 BTeV assumed they could do this in 2009 (though 132 ns not 25 ns) Need pt information ? Magnetic Field in VELO or include other silicon stations

Pile-Up Veto System Two planes of R-measuring sensors Identical to VELO sensors Placed up-stream from interaction point Strips ORed in groups of 4 256 80 Mbit/s links per hybrid Optical links to rad-free barracks FPGA processing performed

Pile-Up Veto: Principle Tracks from same ZPV have the same ratio k RA ZPV - ZA RB ZPV - ZB k  = RA ZPV’ k’ ZB ZA RB ZPV B A k Silicon Sensors (backward!) Histogram combinations Find peaks (Zvtx)  2.8 mm (beam)  53 mm Primary Vertex 1 Primary Vertex 2

Displaced Vertex Trigger Current 2nd Level Trigger algorithm performed in CPU Farm 2D Tracking 2D rz VELO tracking Tracks from beam line form straight lines in rz Primary Vertex search 2D track selection 3D Tracking 3D rfz VELO tracking Add info. From Phi measuring Sensors Only for displaced track candidates 3D confirm L0 m match 3D IP VELO-TT matching Add PT Information Use silicon stations after magnet p, pT estimation L1 decision Total: 1ms not 2s !

System Configuration Outline R projection only Tasks 8 Clustering & Triplet finding & merging Track Identification & filtering In Radiation Free zone 4 sectors per half Track merging Two halfs Vertex identification Impact parameter calculation Final vertex trigger decision 35 processing modules 2 crates 2200 optical links 36 multi ribbon cables (8x8)

Extreme Radiation Environment LHCb VELO will be HOT! Maximum Fluence 1.3x 1014 NIEL 1 MeV neq/cm2/year at 8mm 3.3x 1014 NIEL 1 MeV neq/cm2/year at 5mm 6.6x 1014 NIEL 1MeV neq/cm2/year at 8mm at 1033 Strongly non-uniform dependence on 1/r2 and station (z) Middle station Far station VESPA needs 1015 neq/cm2 charged particle tolerance

Radiation Damage: When replace VELO? Testbeam results, Simulations System components specified to 500V Confident to run to 300V Maintain a reasonable S/N performance 6fb-1 (3-4 years) at 300V

Radiation Hard Technologies Cz, n-on-p, 3D, or pixel technologies – Active R&D, with RD50 Czochralski n-on-p 3D Extreme rad. hard For 4.5 x 1014 24 GeV p/cm2 Depletion voltage = 19V !!!

Strips or Pixels? Both potentially rad. hard 3D, Cz, n-on-p or hybrid pixel detectors Pixels better pattern recognition properties Not major problem (Velo trk eff 97.3%, ghosts 2.3%) But could be trigger advanatge Require approx. 50 m2 pixels Achieve same resolution as Velo Strip geometry more `natural’ Tesselate, strixels But if you can read-out the pixels who cares ? Material Less pixel layers (not R/Phi) Detector, Chip and services (cooling) X0 per layer 1.2% (BTeV), 1.7% (CMS), 1.8% (ATLAS) Thin electronics, typically 500m, achieve 200m BTeV

5mm limit from Accelerator Move Closer 5mm limit from Accelerator Current safe guard ring design 1mm Edgeless technology exits Dope edges etch, laser cut Sensor Design with 5mm active radius Baseline first strip 8mm 7.1mm 10% improvement in IP Impact Parameter Simulation RF foil removed VELO Inner radius 5 mm 36% improvement ! Resolution (microns) 8mm VELO 5mm VELO

Material Budget: RF-foil Total: 19%Xo VELO RF-foil 250m BTeV BTev - 150m thick wires/foil, 6mm from beam In primary vacuum Cryo panels for absorb outgassing TOTEM 1mm from beam (v. diff optics) 150m foil

Prototype in LHC ! Four free station slots available Equipped with cooling tubes Can Prototype in the LHC ! Add extra tracking points, and small amount of extra material

Conclusions Rich Physics Programme for LHCb after 2010 1 st Upgrade at LHC, Increased lumi. Necessitated by radiation environment FPGA based Displaced Vertex Trigger at 1 st Level of Triggering VElo Superior Performance Apparatus Radiation tolerant Si technologies RF foil redesign Reduce material, Move closer to beam Thus the Vespa came to be linked in my eyes with transgression, sin, and even temptation…. And it entered into my imagination not as an object of desire, but as a symbol of an unfulfilled desire." - Umberto Eco