1. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 1 1 Paula Collins (CERN) On behalf of the LHCb and Medipix/Timepix collaborations LHCb Upgrade A Vertex Detector for High Luminosity Upgrading LHCb – our beautiful opportunity Current VELO design Technology options for the upgrade Design concept Strip and pixel options Conclusions Warm up dishes Taste delicious Delicacy from the land Seasonal pickle Can we fulfill our desires? (or should we drink more sake)

2. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 2 2 Mass resolution: 14 MeV Time resolution: 40 fs IP resolution: 14 + 28/pt mm PID: ~ 10 12 bb pairs produced per year Trigger goes from 40→10 MHz in hardware, then to 2kHz in software LHCb: the day 1 experiment bb production correlated and sharply peaked forward-backward Single-arm forward spectrometer : 15-300 mrad σ bb ~ 500 µb in LHCb acceptance Production of B +, B 0, B, b-baryons.., LHCb runs at 50 x lower L by not focusing the beam as much as ATLAS and CMS maximizes probability of a single interaction per crossing LHCb will go immediately to design luminosity K  K : 96.77 ± 0.06%   K : 3.94 ± 0.02% LHCb in numbers 500  b 200  b

3. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 3 Muon HCAL ECAL RICH2 Tracker Magnet TT RICH1 VELO LHCb: in design And in Reality Silicon sensors Vacuum vessel RF box Rectangular bellows Exit window Wake field suppressor Kapton Cables Zoom on the VELO

4. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 4 4 So, now you want more data. why not just turn up L ? LHC is capable of delivering to LHCb ~50% of the L delivered to the GPDs – would be good enough! Going from our current 2.10 32 baseline luminosity to 2.10 33 would only increase the rate of crossings with >= 1 interaction from 10 MHz to 26 MHz This is also compatible with SLHC running in the I high x I low crossing scheme BUT, It’s the trigger! 1MHz read-out rate is the bottle neck in the system  If we increase L we have to increase our E T cut and there is no net efficiency gain! The spillover increases linearly with L and the events become more complex Our current 2.5 us latency is inadequate for a trigger decision of this complexity We are not designed to cope with the additional radiation

5. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 5 5 LHCb upgrade strategy = Move to a full software trigger The idea: Aim to operate at L =2.10 33 Perform entire trigger on CPU farm. The first experiment with no hardware Trigger? The consequence: Read out all sub-systems at 40 MHz Replace all FE-electronics; all silicon modules, RICH-HPDs, FE boards of Calorimeter, Outer Tracker FE Trigger uses ALL event information, reconstructs all primary vertices, and can cut on p T for high i.p. Preliminary studies show that the hadron efficiency will improve by ~2, and will be proportional to L Increase in statistics x20 for hadronic channels and x10 for leptonic channels Of course: we will also improve the detector where we can!

6. VELO: Current Design 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 6  2 retractable detector halves:  7 (37) mm from beam when closed (open)  21 stations per half with an R and  sensor  Operated in secondary vacuum  300μm Al foil separates detector from beam vacuum  Bi-phase CO 2 cooling system 6 RF foil 21 +2 modules Base Kapton cables Cooling manifold 6

7. VELO sensors 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 7 7 Ф measuring sensors:  2 regions  Short/long strips  Pitch=36-97μm  Stereo angle. 7 silicon edge just 7 mm from beam! R measuring sensors:  45 o quadrants  Pitch=40-102μm 512 strips 512 strips 512 strips 512 strips 683 inner strips 1385 outer strips 300  m n in n strip sensors (Micron Semiconductor) Double metal layer for signal routing

8. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 8 8 Operating up to ~120 fb -1 Flux: 0.8x10 14 n eq cm -2 per fb -1 TID (Electronics): 3.7 MRad per fb -1 at tip = 7mm 8 Upgrading the VELO: Irradiation issues 500 50 5 Radius (cm) Dose after 100 fb -1 n eq cm -2 x 10 16 TID (MRad) After this dose @ 900V we expect 102 uA / cm -2 at -25 o C CCE of ~ 8.5 ke - Thermal runaway at the tip is the issue tip of current VELO T. Affolder TIPP 09

9. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 9 9 Upgrading the VELO: Bandwidth The occupancy is relatively low, but at 40 MHz the data rates are enormous Rough estimate gives 5 Gbit s -1 per FE for current design Minimum of 1400 optical links needed for whole detector Clustering issues become very important for getting the data off-chip 9 Radius (cm) Particle Hits / Event / cm 2 Averaged over all sensors L = 2.10 32 L = 5.10 32 L = 10.10 32 L = 20.10 32 a  b  Dedicated R&D needed on High speed copper cables within tank Feedthroughs/PCBs

10. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 10 Radius of first measured point (mm) Original layout Effect of removing 10 stations The global constraint is the ~1.2 m length of the vacuum tank From our previous studies we know that the impact parameter resolution is well described to first order by 10 Upgrading the VELO: design issues Keep down material before the second measured point High momentum tracks: keep the first measured point as precise as possible (i.e. fine pitch and good position information Keep the extrapolation distance as small as possible: Many stations Curved inner shape Minimum inner radius Edgeless technology an advantage Remember though the radiation at the tip Behaviour of impact parameter resolution as VELO material is increased Behaviour of impact parameter resolution as foil material is decreased

11. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 11 Possible upgraded Strip layout 200  m Diamond heat plane routing out Beetle signals 20 Beetle40 chips Inner radius at 7.5mm, 25- 30  m pitch (edgeless sensors would bring 10% improvement) 200  m thin silicon sensor Radiation length 0.6%; 0.4% for first mm Double sided module Cooling channel Sensor would be p-stop or p-spray. Hammamatsu are interested in propotyping

12. 12 FPIX2: Bringing BTeV experience to LHCb Precision X sensor Front View Back View Precision Y sensor TPG or Diamond Substrate 50x400 um pixels in double layer Staggered Pixel ladders (no gaps) Heat carried off via CF-encapsulated TPG, through flexible PGS to LN 2 heat sink Pixel solution could help in pattern recognition and processing time Material estimate could be pushed to 1.5% per station plus ground / power planes

13. Jianchun Wang 13 FPIX in action at Fermilab testbeam 2009 VELO Pixel X/Y Pixel Y ~ 1 m

14. Jianchun Wang 14 Pixel Detector Pixel telescope aperture: ~ 35 mm x 35 mm, 4/5 planes. FPIX2 readout chip (128 rows x 22 columns ). Pixel size 50  m x 400  m (600  m for first and last columns). Only binary information used (FPIX2 has 3-bit ADCs). Number of Rows Number of Hits (arb Unit) Residual (mm)  = 11.3  m 75.8% Beam Monitoring

15. Next steps for FPIX 15 Start by bench test of FPIX2 to understand factors to be improved (analog performance, readout speed…) Design and prototype a modified version to address some of the most urgent problems [bco, readout clock speed…] in collaboration with Fermilab Migration to 0.13  m technology if this solution is chosen, merge with other project(s) for final system?

16. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 16 TIMEPIX solution Pixel readout chip based on Medipix2 256 x 256 pixels 55  m square, and chip is 3 side buttable. Single Sided Modules Possible!! (X 0 ) By using TSV (through silicon vias) dead side can be reduced to 0.8 mm in Medipix3 (out of 1.5 mm) Analogue power consumption 6  W per pixel. TOT provides better than 6 bit equivalent ADC resolution Upgrade being considered to 90 nm technology (power consumption, density of logic and radiation hardness benefits) VELOPIX derivative options to be included in next design cycle (2009/2010) of TIMEPIX2, derived from MEDIPIX3 (delivered early 2009) in 130 nm 14100 800 Medipix3 Chip dimensions

17. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 17 Introducing the TIMEPIX for HEP Triggered by the need of gaseous detectors to measure the arrival time of the electron in the sensitive gas volume  3D (position + time) !!! The Medipix2, originally based on shutter mode and developed for imaging, was redesigned to incorporate a time stamp with a tunable resolution of 100 to 10ns. Requirements: Keep Timepix as similar as possible to Medipix2 in order to benefit from large prior effort in R/O hardware and software Avoid major changes in pixel and/or readout logic – risk of chip failure due to poor mixed mode modeling Add possibility of programming pixel by pixel arrival time or TOT information DESY testbeam in November 2006 (A.Bamberger et al)

18. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 18 Time Over Threshold principle similar principle as ATLAS pixel, Panda (Topix)… Preamp has fast rise (90ns) but slow (500ns -2500ns) constant-current-return to zero. (programmable Ikrum) ToT=f(Q, threshold) E.g. 22ke- ~1us= 40 x 25ns 4 bit threshold: RMS of 35 e after adjustment Residual errors come from non linearities for small charge and slow return to baseline Comparator output Comparator threshold LE TE ToT = TE - LE Q TOT calibration Before After

19. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 19 Potential module design Silicon (1-3 pieces) 55x55 mm pixels + 800 mm pixels in areas under chip periphery 10 Tiimepix chips (periphery indicated in white) Ground plane (aluminised directly onto diamond) Diamond thermal plane with cutouts immediately above TSV regions Power strips and signal routing area Cooling channel Cross section cutout region

20. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 20 Thermal performance (preliminary study) 200 um CVD diamond (k=1600 W/m K) 10 ASICS (150um, 1W, 14x14mm^2) Access windows in diamond plane Sensor (150um) Opening 7x14 mm^2 78 mm 47 mm coolingBack+ edges back onlyedges only back + edges Diamond planes 1112  T to tip 4.1 o 5.3 o 9.3 o 2.0 o

21. 21 Medipix3 X-ray Irradiation Total ionizing dose measurement on a single chip up to 400Mrad (continuously over 4 days) Confirmation of previous single transistor studies on 130nm CMOS Chip readout DACs, LVDS etc remained operational for full dose. A design issue means a single Pixel DAC is particularly sensitive to damage, causing a change in operating point after 1Mrad. This will be fixed in next version.

22. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 22 Next steps for LHCb Demonstrate the ability of the chip to reconstruct landaus and perform charged particle tracking (see next slides) Investigate methods of moving data off chip at required speeds. Studies have started, digital processing within pixel array may be necessary Check that the timewalk, which could potentially affect LHCb physics, is controllable Testbench features of baseline module 31/08/09 22 25 ns

23. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 23 July-August 2009: Dedicated Timepix Telescope A telescope was constructed with 6 double angled (at 9 o ) planes. 4 Timepix and 2 Medipix sensors were used. The DUT (in this case another Timepix chip) position and angle can be controlled by a stepper motor to reduce the number of interventions Thanks to our colleagues from the CMS SiBT and the EUDET-Spider collaboration who enabled us to run in the H6 and H8 120 GeV pion beams 6 plane timepix/Medipix telescope DUT

24. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 24

25. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 25 TIMEPIX example data with mips Mean value of cluster size varies from 2.5-3.5 between 0-18 o Landau very clean; noise is negligible Precise tracking clearly picks out binary and charge sharing regions Landau peak, Timepix at 0 o Cluster size, Timepix at 10 o Track residuals Timepix at 6 o Example SPS Beam Profile in TIMEPIX chip

26. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 26 TIMEPIX and MEDIPIX charge sharing Charge Sharing Y axis: Track prediction X axis: cluster measurement (simple charge weighting) 9 degree angle10 degree angle6 degree angle MEDIPIX (binary) 10 o TIMEPIX (TOT)

27. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 27 Preliminary result: TIMEPIX resolution Estimated track contribution to residual Timepix: 55  m pitch Unbiased x Residual as a function of angle Rotation about Y axis PRELIMINARY: UNCALIBRATED DATA!!

28. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 28 Preliminary result: HV scans at angle Width of y residual Scanning the HV up and then down again shows a clear shift in the residual means for highly angled tracks 10 degrees 18 degrees Mean of x residual Mean of y residual (rotation is about y axis).. 18 degrees Width of x residual The residual width also shows a clear dependence on orientation. This is as expected, and illustrates the sensitivity of the measurement

29. 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 29 Other measurements on the way Resolution as a function of  threshold  2d rotation  Number of bits used Efficiency and Noise study (already rather clear) Pulse shape reconstructed directly from ToT data 3d sensor Etc.

30. VELO Upgrade – Yes We Can! The LHCb upgrade can go ahead independently of SLHC  X 10 statistics for leptonic channels  X 20 statistics for leptonic channels Reading out the VELO for 100 fb -1 at 40 MHz requires a major redesign Options considered are strips and pixels (FPIX flavour, as demonstrated for BTeV, or Timepix flavour – new concept) For the first time, the use of Timepix has been demonstrated to be extremely effective for charged particle tracking A Timepix based telescope has produced track extrapolation precisions of around 2.5 micron and shown extremely stable and noise free performance. Can be upgraded to the new Medipix quad readout (significantly reduced dead time) and to include track time stamping R&D is going ahead to adapt the Timepix to LHCb requirements 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 30

31. 31 Medipix3 X-ray Irradiation Accumulated dose of 396 Mrad 9mV NMOS target DACs (Preamp) Reproduced with thanks from presentation by X. Llopart

32. Timepix cluster sizes 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 32 0o0o 8o8o 2o2o 10 o 12 o 18 o

33. Example Track Fit Residuals for TIMEPIX only 31/08/09 LHCb Upgrade – Paula Collins 7 th International Hiroshima Symposium 33 TIMEPIX, In track fit TIMEPIX DUT, Not in track fit TIMEPIX, In track fit TIMEPIX, In track fit TIMEPIX, In track fit MEDIPIX, Not in track fit MEDIPIX, Not in track fit 0 mm 470 mm 310 mm 235 mm 160 mm 390 mm 80 mm The numbers in mm refer to the z position of the stations