Performance of a double-metal n-on-n and a Czochralski silicon strip detector read out at LHC speeds Juan Palacios, On behalf of the LHCb VELO group J.P. Palacios, Liverpool
Outline LHCb and VELO performance requirements Performance of prototype sensor/chip Si sensor / readout chip design Autumn 2003 test beam Performance results Conclusions Future technology: Czochralski silicon Motivation Test sensor / test beam setup Results Conclusions J.P. Palacios, Liverpool 2
The LHCb experiment A B-physics CP violation experiment Good primary, secondary vertex resolution Good particle ID A small angle forward spectrometer with excellent PV and IP resolution HCAL TT1 ECAL Muons VELO RICH1 Tracker RICH2 J.P. Palacios, Liverpool 3
Requirements to the VELO (1) As vertex detector Get as close to beam as possible Move sensors to within 7mm of beamline Highest resolution on first measured point Sensor resolution, material before 1 st hit As part of trigger Fast 2D and 3D tracking Rφz geometry for pattern recognition Fast readout + high S/N, overspill < 30% OPEN (injection) CLOSED (physics) J.P. Palacios, Liverpool 4
VELO requirements (2) VELO as tracker Cluster resolution starts at 5µm 21 Rφ measuring planes keep material in acceptance as low as possible VELO track VELO seeds LHCb tracks Upstream track Long track (forward) Long track (matched) T track T seeds Downstream track Number of hits per particle 40 20 Long tracks Downstream tracks Upstream tracks T tracks VELO tracks highest quality for physics (good IP & p resolution) needed for efficient K S finding (good p resolution) lower p, worse p resolution, but useful for RICH1 pattern recog useful for RICH2 pattern recognition useful for primary vertex reconstruction (good IP resolution) 0 0 2 4 6 Pseudorapidity J.P. Palacios, Liverpool 5
Fluence nearest IP (n eq cm -2 year -1 ) VELO requirements (3) Harsh, inhomogenous radiation environment After irradiation V Dep varies across sensor Flux 5x10 12 to 1.3x10 14 n eq cm -2 yr -1 depending on r and z V dep after 1 year Need good resolution, efficiency, S/N, low material, radiation hardness and 40MHz readout for at least 2 years of LHC operation THIS IS CHALLENGING! J.P. Palacios, Liverpool 6
VELO sensor/chip solution Sensor n-on-n technology Oxygenated Thin (200-300µm) Fine pitch (40-100µm) R and Φ geometries Double metal Beetle Readout chip Deep sub-micron rad-hard technology 40MHz LHC readout speed 4 output (32 channel) mode Highly configurable Large parameter space Found shaper feedback voltage (V fs ) to be most relevant cf. pulse height and overspill Φ measuring Readout chips Chips R measuring diodes routing Routing lines lines J.P. Palacios, Liverpool 7
R measuring prototypes PR03 pre-production prototype Close to final design Several fabricated Tested in beam PR04 production prototype Minor geometry changes First sensors May 2004 J.P. Palacios, Liverpool 8
Autumn 2003 testbeam Objective: test 200µm PR03/Beetle1.2 Sensor/chip closest to final design LHC conditions: 4 analogue output 40 MHz First beam test of 200µm n-on-n double metal sensor at LHC speeds Check fulfilment of VELO requirements Signal, S/N Signal remainder after 25ns (overspill) Use tracks to investigate full pulse shape Exploit configurability of Beetle chip Find best set of running parameters J.P. Palacios, Liverpool 9
VELO test beam setup CERN PS X7 Test beam 120 GeV pions Telescope to reconstruct tracks Made from earlier prototypes: Rφ geometry Beetle1.2 on PCB board Readout region: 45 o strips, average pitch 44µm 200µm n-on-n Micron PR03 R sensor J.P. Palacios, Liverpool 10
Signal estimation Events 400 350 300 250 200 150 100 50 Landau dominated 0-40 -20 0 20 40 60 80 100 Integrated charge 4 strips Integrated charge Events 1200 Track Charge integration strips Good strips Other strips Estimate charge by integrating signal on strips neighbouring track extrapolation point. 1000 800 600 400 200 0 Gaussian dominated -40-20 0 20 40 60 80 100 Integrated charge 4 strips Fit Landau*Gaussian to signal and signal/noise distributions J.P. Palacios, Liverpool 11
ADC pulse shapes Charge / ADC counts 35 30 25 20 15 Vfs=1000mV Vfs=500mV Vfs=300mV Vfs=100mV Vfs=0mV Charge / ADC counts 35 30 25 20 10 15 5 10 0 5-5 Zoom 0 10 20 30 40 50 60 70 80 90 100 25 30 35 40 45 50 55 TDC/ns TDC/ns Signal pulse shapes for different values of V vf (shaper feedback voltage) J.P. Palacios, Liverpool 12
S/N curves Signal / Noise 14 12 10 8 Vfs=1000mV Vfs=500mV Vfs=300mV Vfs=100mV Vfs=0mV 12 10 6 4 2 8 6 Zoom 0-2 4 0 10 20 30 40 50 60 70 80 90 100 25 30 35 40 45 50 TDC/ns TDC/ns S/N pulse shapes for different values of V vf (shaper feedback voltage) J.P. Palacios, Liverpool 13
Summary of results Signal/Noise 20 15 10 5 LHCb starting requirement Trigger minimum requirement Signal amplitude/adc 60 50 40 30 20 10 0-200 0 200 400 600 800 1000 Vfs/mV 0 0 200 400 600 800 1000 200µm data 300µm extrapolation Vfs/mV Overspill / Peak Signal 0.5 0.4 0.3 0.2 Trigger maximum overspill 30% Thin n-on-n double metal sensors are challenging! 0.1 0 0 200 400 600 800 1000 Vfs/mV J.P. Palacios, Liverpool 14
Undershoot region Charge / ADC counts 30 25 20 15 10 5 0-5 -10 0 20 40 60 80 100 120 140 TDC/ns Overspill / Peak Signal 1.2 1 0.8 0.6 0.4 0.2 0-0.2-0.4 0 20 40 60 80 100 TDC-TDC(peak) / ns Undershoot remains at 10% level 100ns after peak. Thought not to be a problem given the VELO occupancy J.P. Palacios, Liverpool 15
PR03/Beetle1.2 conclusions The first beam test of a 200µm n-on-n R measuring sensor read out at 40MHz through a Beetle1.2 chip has been carried out successfully The S, S/N and overspill characteristics for a set of readout chip parameters have been measured using 120GeV pions With this technology 300µm R measuring sensors are desirable for LHCb J.P. Palacios, Liverpool 16
Future technologies A possible upgrade of the VELO detector would provide a great opportunity to try new sensor technologies Czochralski silicon could be an alternative to float zone Czochralski is the standard, cheaper, mono-crystaline silicon used by most IC manufacturers P-bulk float zone another interesting option see poster by Gianluigi Casse J.P. Palacios, Liverpool 17
Czochralski test beam Why Czochralski? High resistivity (~1KΩcm) now available Very high oxygenation level RADIATION HARD Standard silicon 10 15 cm -3, DOFZ 10 17 cm -3, Cz 10 18 cm -3 Low cost Could be significant for very large scale future sensors Depletion voltage vs. fluence Some evidence that it is more stable Some evidence that it might not type-invert J.P. Palacios, Liverpool 18
Cz test sensor Sensor produced at the Helsinki Institute of Physics. THANKS!!! First large scale prototype sensor! High resistivity (1150Ωcm) p-on-n strip sensor 50µm pitch, 380µm thick, 6.2 cm strips Can a full-scale prototype be irradiated and tested at LHC speeds? Sensor bonded to 3 SCTA 40MHz readout chips (384 strips) In test beam summer 2002 Irradiated to 7.8x10 14 protons/cm 2 In test beam summer 2003 J.P. Palacios, Liverpool 19
Cz 2003 test beam analysis Sensor non-uniformly irradiated Irradiation level along sensor measured All data samples separated into 3 groups according to fluence 7x10 14 p/cm 2 ~3 years 4.25x10 14 p/cm 2 ~2 years 1.25x10 14 p/cm 2 ~0.5 years J.P. Palacios, Liverpool 20
Cz 2003 test beam analysis Signal estimation same as previous analysis Obtain signal level as function of irradiation and bias voltage 98.84+1.66 60.33+1.20 34.43+0.74 Φ=1.25x10 14 p/cm 2 Φ=4.25x10 14 p/cm 2 Φ=7x10 14 p/cm 2 Landau*Gaussian fits for different irradiation levels, V bias =167V. J.P. Palacios, Liverpool 21
Cz 2003 test beam results 1.25x10 14 p/cm 2 4.25x10 14 p/cm 2 7x10 14 p/cm 2 PRELIMINARY!!! J.P. Palacios, Liverpool 22
Cz test beam conclusions Clear loss of charge collection efficiency with irradiation Can be recovered by applying higher bias voltages So far seems similar to float zone Si Analysis to determine if it could gain a year under LHC operation underway Non-irradiated data samples Higher bias voltage points But this is the first such sensor ever built and read out with fast electronics so we are optimistic for the future! J.P. Palacios, Liverpool 23