The LHCb VELO and its use in the trigger Thomas Ruf Vertex 2001 23-28 September 2001 Introduction Silicon R&D Second Level trigger Pile up VETO trigger

Introduction VErtex LOcator of the LHCb experiment LHCb: Systematic studies of CP in the beauty sector by measuring particle - antiparticle time dependent decay rate asymmetries. LHCb requires: Reconstruction of pp-interaction point Reconstruction of decay vertex of beauty and charm hadrons Standalone and fast track reconstruction in second Level trigger (L1)

VELO Overview Vacuum Vessel ~1 m Introduction  silicon sensors are placed in vacuum Vacuum Vessel Precise vertexing requires, to be: as close as possible to the decay vertices with a minimum amount of material between the first measured point and the vertex Number of silicon sensors: 104 Area of silicon: 0.32m2 Number of channels: 204,800 ~1 m

VELO Setup One detector half: p p forward Introduction Positioning and number of stations is defined by the LHCb forward angular coverage of 15mrad <  < 390mrad together with the inner/outer sensor dimensions Also need to account for spread of interaction region: s = 5.3 cm partial backward coverage for improved primary vertex measurement One detector half: Interaction region p p forward Final configuration: 25 stations 1 station = 2 modules (left and right) 1 module = 2 sensors

Introduction Radiation Sensors have to work in a harsh radiation environment: max. fluences: 0.5 x 1014 - 1.3 x 1014 neq / cm2 / year N*r-a a = 1.6  2.1 neq = damage in silicon equivalent to neutrons of 1 MeV kinetic energy

Introduction Sensor Design Azimuthal symmetry of the events suggests sensors which measure f and R coordinates. Advantages of Rf geometry: Resolution: Smallest strip pitch where it is needed, optimizing costs/resolution. Radiation: Short strips, low noise, (strixels, 40mm x 6283mm) close to beam L1: Segmented R-sensor (45o) information is enough for primary vertex reconstruction and impact parameter measurement. Inner radius is defined by the closest possible approach of any material to the beam: 8 mm (sensitive area)  LHC machine Outer radius is constrained by the practical wafer size: 42 mm

Sensor Design Strips are readout by using a double metal layer Introduction Sensor Design Strips are readout by using a double metal layer Analog readout for better hit resolution and monitoring 2048 strips / sensor  16 FE chips

Silicon R&D Tested Prototypes VELO Design Challenges Varying strip lengths Double metal layer Regions of fine pitch Large and non-uniform irradiation VELO Technology Choices Thickness Oxygenation Cryogenic Operation Segmentation p or n strips ? Tested Prototypes Double sided DELPHI sensor thickness: 310 m

Silicon R&D Tested Prototypes VELO Design Challenges Varying strip lengths Double metal layer Regions of fine pitch Large and non-uniform irradiation VELO Technology Choices Thickness Oxygenation Cryogenic Operation Segmentation p or n strips ? Tested Prototypes Hamamatsu n-on-n thickness: 300 m

Silicon R&D Tested Prototypes VELO Design Challenges Varying strip lengths Double metal layer Regions of fine pitch Large and non-uniform irradiation VELO Technology Options Thickness Oxygenation Cryogenic Operation Segmentation p or n strips ? Tested Prototypes MICRON p-on-n thickness: 200/300 m

Results about irradiated sensors Silicon R&D R&D Results ~5% of VELO sensors tested with beam First confrontation with alignment issues 40MHz readout chip SCT128A Resolution vs angle and pitch See talk of M. Charles Best resolution: 3.6 m Trigger performance simulation http://lhcb-vd.web.cern.ch/ lhcb-vd/TDR/TDR_link.htm May 2001 Results about irradiated sensors

Sensor readout with 25ns electronics (SCT128A) Silicon R&D n-on-n Prototypes Variable irradiation with 24 GeV protons at the CERN-PS Sensor readout with 25ns electronics (SCT128A) Repeater card S/N = 21.5 3-chip hybrids Most irradiated region corresponds to 2 years of LHCb operation for the innermost sensor part. Temperature probes

Silicon R&D p-on-n Prototypes After irradiation, silicon needs to be fully depleted, otherwise: Resolution degrades Charge Collection Efficiency Resolution [mm]

Silicon R&D p-on-n Prototypes After irradiation, silicon needs to be fully depleted, otherwise: Charge is lost to double metal layer 2nd metal layer Size of boxes proportional to CCE

VELO Technology Choices Silicon R&D VELO Technology Choices n-strips, safest solution, sensors can be operated underdepleted Thickness = 300 m, OK for radiation hardness and material budget Oxygenation: nice to have, but not mandatory, less of interest for p-on-n Fully depleted for >2 years, Ubias < 400 V Prototype n-on-n sensors even for 4 years With n-on-n, can accept 40% under-depletion  extending the lifetime even further Operation model: 100 days constant fluence, T=-50C 14 days at +22oC Prototype n-on-n sensors behaved much better than expected for “standard” silicon

LHCb Trigger System Data is kept in analog/digital pipelines L0: look for high pt particles hadron (HCAL), muon (MUON) or electron (ECAL) pile up VETO 40 MHz  1 MHz Fixed latency: 4.0 ms p L1: vertex trigger 1 MHz  40 kHz L1 buffer = 1927 events L0 YES: Data is digitized, but stays in memory of individual detector electronics L1 YES: Digitized data is read-out, event building  CPU farm L2: refined vertex trigger 40 kHz  5 kHz L3: online event selection 5 kHz  200 Hz

L1 (Vertex) Trigger Input rate: 1 MHz Output rate: 40 kHz LHCb Trigger L1 (Vertex) Trigger Input rate: 1 MHz Output rate: 40 kHz Maximum latency: ~2 ms Purpose: Select events with detached secondary vertices Needs: Standalone tracking and vertex reconstruction ~ 1.5 tracks in the seeding region (innermost 450 sectors of R-sensors)  track reconstruction is straightforward Occupancy < 1% per channel

Algorithm 2d track reconstruction using only R-hits Primary vertex: L1 Vertex Trigger Algorithm r 2d track reconstruction using only R-hits Efficiency: 98% z triplets Primary vertex: Histogram of 2-track crossings  z F-segmentation  x,y sz,y  70 mm sx,y  20 mm Impact Parameter Reject events with multiple interactions Test beam 3d reconstruction for tracks with large impact parameter Efficiency: 95% Search for 2 tracks close in space Calculate probability to be a non b-event based on impact parameter and geometry

Implementation Basic Idea: 150-250 processors SWITCH CPU farm L1 Vertex Trigger Implementation Readout Unit 150-250 processors Basic Idea: SWITCH CPU farm Digitizer Board: zero-suppression, common mode correction, cluster finding 1 MHz x100 x24 2d torus topology based on SCI shared memory Challenge: 4 Gb/s and small event fragments of ~170bytes 680 Mb/s, limited by number of senders

Execution Time of Algorithm L1 Vertex Trigger Execution Time of Algorithm 450 MHz Pentium III running Windows NT Bp+p- Minimum bias L1 latency Expect CPU’s to be 10 times faster in 2005

Minimum bias retension L1 Vertex Trigger Physics Performance Technical Proposal With new Event Generator, performance degraded by a factor of 2 ! Minimum bias retension Main limitations: No momentum information significance of impact parameter No particle ID Working point 40kHz Signal efficiency Link L0 objects: large pt (e, m, h) with VELO tracks. Investigate effect of possible momentum information. Recent developments:

Super Level 1 Example: Matching with HCAL clusters L1 Trigger New ideas Super Level 1 matching efficiency Bp+p- : 78% for one p BJ/y(mm)Ks: 96% for one m HCAL MUON ECAL Example: Matching with HCAL clusters B dl = 4 Tm, pt kick = 1.2 GeV/c New Bp+p- BJ/y(mm)Ks Bp+p- TP BJ/y(mm)Ks L1 rate In general, gain back factor 2, in some channels even more.

Mini Level 1 For final answer: See Trigger TDR in 2002 L1 Trigger New ideas Mini Level 1 Silicon tracking station 30% additional data to be send to L1 farm momentum resolution: s(pt)/pt  20%  B0s  Ds-(K+K--) K+  B0d  +- For final answer: See Trigger TDR in 2002

LHCb chooses luminosity for maximum number of single interactions LHCb Trigger L0 Pile Up VETO Input rate: 40 MHz Output rate: 1 MHz Latency: 4.0 ms Purpose: Remove events with multiple interactions. Why ? Multiple interactions are more difficult to reconstruct (specially for L1) and fill bandwidth of L0 (~ 2x probability to pass L0). b-event rate: 25kHz Number of inelastic interactions/bunch crossing as a function of luminosity LHCb chooses luminosity for maximum number of single interactions Cross-section Luminosity Bunch crossing frequency  30 MHz with At L=2x1032cm-2s-1: P>1/ P1  24% b-events: P>1/ P1  41%

Concept ZB ZA RB RA ZPV 2 silicon R-stations upstream of VELO B A L0 Pile Up VETO Concept RB [cm] RA [cm] True combinations All combinations ZPV [cm] ZB ZA RB RA ZPV 2 silicon R-stations upstream of VELO B A If hits are from the same track:  ZPV mask hits belonging to P1 search next highest peak P2 (peak size S2) classify: if S2<Slimit then single, else multiple build a ZPV histogram search highest peak P1

Implementation 2 R-stations upstream of the VELO L0 Pile Up VETO Implementation 2 R-stations upstream of the VELO OR of 4 channels, binary readout, 80Mbit LVDS, 512 lines Frontend chip: BEETLE running in comparator mode Large FPGA: >350k gates and >600 I/O pins. Candidates: Altera EP20K400, XILINX XC4000, … Performance: Gain of 30-40% of single bb-events at optimal luminosity

Summary Technical Design Report of the LHCb VErtex LOcator is completed: n-on-n silicon strip sensors are the baseline. Prototyping with different companies continues. The VELO plays an important role in the second level trigger: Standalone track finding, primary vertex reconstruction, impact parameter determination Silicon sensors are also used in the first level trigger for a fast determination of the number of interactions.