1. The DØ Silicon Microstrip Tracker Frank Filthaut University of Nijmegen / NIKHEF NIKHEF, 4 August 2000

2. 4 August 2000NIKHEF Tevatron Run II Upgrade Modification of Tevatron parameters: Start of Run II: 1 March 2001 Aim: collect 2 fb -1 in  two years (might be more…) Switch from 396 ns to 132 ns bunch spacing at luminosity  10 32 cm -2 s -1 Keep zero crossing angle for 396 ns operation; aim for 136  rad angle for 132 ns operation Beam spot:  35  m transverse, 25 cm longitudinal (10 cm for nonzero crossing angle) Run 1Run 2 (initial) Run 2 (eventual) Bunch spacing (ns) 3500396132 Luminosity (10 31 cm -2 s -1 ) 0.168.621 Interactions/ crossing 2.52.31.9

3. 4 August 2000NIKHEF DØ Run II Upgrade p Addition of central axial 2T magnetic field (SC solenoid in front of calorimeter cryostat) p Replacement of tracking system by combination of scintillating fibers (Central Fiber Tracker) and silicon sensors (Silicon Microstrip Tracker) p Central (CPS) and forward (FPS) preshower detectors p Extend muon chamber coverage to larger , smaller granularity p Upgraded calorimeter, trigger, DAQ electronics Physics aims: p B-tagging based on b lifetime p Improved electron and muon identification and triggering p Improved tau identification p Charge sign determination High p t central physics (tt, EW, Higgs and other searches): high multiplicity Low p t physics (bb, QCD): requires good forward coverage (e.g. lepton ID for |  | < 3)  

4. 4 August 2000NIKHEF DØ Run II Upgrade

5. 4 August 2000NIKHEF DØ Run II Upgrade - Tracking Forward Preshower Silicon Microstrip Tracker Fiber Tracker Solenoid Central Preshower All share the same SVX IIe front-end electronics

6. 4 August 2000NIKHEF 1 2 3 4 5 6 1 2 3 4 5 6 7 8 9 10 11 12 SMT Design Basic SMT Design: BarrelsF-DisksH-Disks Layers/planes4124 Readout Length 12.4 cm7.5 cm14.6 cm Inner Radius2.7 cm2.6 cm9.5 cm Outer Radius9.4 cm10.5 cm26 cm p 6 barrels p 12 F disks p 4 H disks Axial strips to be used in 2 nd level Silicon Track Trigger (STT)  stringent requirements on alignment Totals: 793k channels 3.0 m 2 (of which 1.6 m 2 DS)

7. 4 August 2000NIKHEF SMT Design p Layers 1 (3): 12 (24) DS, DM 90 0 ladders produced from 6” wafers (barrels 1 & 6: SS axial ladders from 4” wafers) p Layers 2 (4): 12 (24) DS 2 0 ladders produced from 4” wafers SMT barrel cross-section: Ladder count: 72 SS + 144 DS (90 0 ) + 216 DS (2 0 )

8. 4 August 2000NIKHEF Anatomy of a Ladder p Ladders supported by “active” (cooled) and “passive” bulkheads p Ladders fixed by engaging precision notches in beryllium substrates on posts on bulkheads p Beryllium cools electronics  expect chips to operate at 25 0 C using 80% H 2 O/20% ethyl glycol mixture at –10 0 C  silicon should be at 5-10 0 C p High Density Interconnect (HDI) tail routed out between outer layers p Carbon-fibre/Rohacell rails glued to sensors for structural stiffness

9. 4 August 2000NIKHEFSilicon p Significant fraction of silicon will undergo type inversion p Prefer high initial V depl for silicon at low radii  Specification of V depl, V break  Selection of sensors at hand Calculated radiation dose as a function of radius (z=0): Quality control at vendors & DØ institutes: p I leak < 10  A at V bias = V depl + 10V p 20V < V depl < 60V p Polysilicon bias resistors: 1 M  < R bias < 10 M  (DC bias pad) p R inter-strip  G  p Coupling capacitors: 10-15 pF/cm p f(shorted cap’s) < 2% at min(V depl +15V, 90V)

10. 4 August 2000NIKHEFSensors p Need 144 sensors (2/detector) p Pitch: 50  m p Typical V depl : 30 V p All sensors delivered by Micron p Flatness problems (  60  m) a concern Single-sided

11. 4 August 2000NIKHEFSensors p Need 432 sensors (2/detector) p Pitch: p-side 50  m, n-side 62.5  m p Typical V depl : 20-40 V p Slow sensor delivery by Micron  accepting sensors with higher R bias Double-sided 2 0 p-siden-side

12. 4 August 2000NIKHEFSensors p Need 144 sensors p Pitch: p-side 50  m, 90 0 n-side 153  m p Produced from single sensor (6” technology) p Using DM layer, gang 2 n-side strips together to form 1 readout channel p Typical V depl : 50 V Double-sided, double-metal

13. 4 August 2000NIKHEFSensors p Sensor delivery from Micron has been slow (30% yield) mainly due to p-stop defects on mask (noise affecting  10-15 strips)   5 sensors / week  schedule problem, accepting few sensors with 1 p-stop defect Double-sided, double-metal

14. 4 August 2000NIKHEFSensors p Need 144 sensors (12 detectors/disk) p Pitch: n-side 62.5  m, p-side 50  m (flexible pitch adaptor on n-side) p Stereo angles:  15 0 p Sensor delivery:  Micron: delivered 125 sensors with reasonable characteristics: V depl  60-70 V  Eurisys: delivered 65 sensors: First two batches: V depl  15-20 V, V break  70-80 V Last batch (  25 sensors) with better implants: V depl  30-40 V, V break  80-100 V F-wedge

15. 4 August 2000NIKHEFSensors p Need 384 sensors (2/detector, 48 detectors/disk) p Single-sided, glued back-to-back p Pitch: 80  m p Stereo angles:  7.5 0 p Typical V depl : 60 V p All sensors delivered by ELMA H-wedge

16. 4 August 2000NIKHEF SVX IIe Chip p SVX chip originally designed by LBL - FNAL for readout of CDF vertex detector, optimised for capacitances of 10-35 pF, ENC = 350e - + 50e - /pF p 128-channel 8-bit digital chip, 1.2  m rad-hard technology p Both signal polarities p Rise time set to integrate 99% of signal in 100 ns (for 132 ns operation) p Double correlated sampling for dynamic “pedestal” subtraction p Front end capacitor discharged during “beam gaps” p Pipeline depth 32 max. p 106 MHz readout speed (both edges of 53 MHz clock)

17. 4 August 2000NIKHEF SVX IIe Chip p Daisy-chained readout of max. 9 chips p Online sparsification using common threshold. Modes:  All channels  Only above threshold  Include 1 or 2 neighbours on either side (even if on adjacent chips) p Adjustable ramp rate (dynamic range) p Power dissipation  5 mW/channel p Unlike SVX III now in use by CDF, not deadtime-less

18. 4 August 2000NIKHEF High Density Interconnect p Two-layer flex-circuit mounted directly on silicon, housing SVX chips as well as passive electronics p Kapton based, trace pitch 200  m p Connects to “low-mass” cable using Hirose connector p 9 different types for the 5 sensor types  2 for each sensor type except H disks  2 types for each ladder differ only in tail length p Laminated to beryllium substrate (total mass  0.041 X 0, of which 0.014 X 0 from Si) Need 912 HDI’s 9-chip HDIH-wedge HDI

19. 4 August 2000NIKHEF Production Sequence Probe Test Silicon Sensor at Micron Silicon Sensors Test Bare HDI Laminate HDI Mount components on HDI Test stuffed HDI Burn-in HDI HDI Build Ladder/Wedge Mount HDI on Silicon Wirebond Detector Burn-in Ladder Laser Test Build Full Wedge (H) Mount on Support structure Read out Detector Probe Test Silicon Sensor in house “fail” Outside Co. University Fermilab Micron Eurisys ELMA Promex Silitronics Dyconex

20. 4 August 2000NIKHEF Ladder Production in steps (9-chip) 1. Apply pattern of non-conductive epoxy on p-side beryllium 2. Align beryllium with respect to active sensor, apply pressure and cure for 24 hr 3. Align active & passive sensors w.r.t. each other, apply wirebonds. Then use separate fixture to position carbon-fibre rails. Use conductive epoxy to ground “passive” beryllium. Cure for 24 hr

21. 4 August 2000NIKHEF Ladder Production in steps (9-chip) 4. Use “flip fixture” to have n-side on top 5. Apply epoxy to n-side beryllium, fold over and secure HDI. Apply pressure and cure for 24 hr. Then apply n-side Si-Si and Si-SVX wirebonds 6. Encapsulate bonds at HDI edges. Connect “active” beryllium to cable ground

22. 4 August 2000NIKHEF Testing & Repairs Bonds need to be plucked Bad ground connection p Broken capacitors: cause SVX front-end to saturate, tends to affect neighbouring channels as well  pluck corresponding bonds p Bad grounding of beryllium substrates causes large pedestal structures as well as high noise  ensure R Be-gnd < 10  p Repair broken / wrong bonds p Replace chips / repair tails damaged during processing

23. 4 August 2000NIKHEF Burn-in & Laser Tests Dead Channel Laser Laser Test: p Energy just < Si bandgap (atten. length  400  m  test whole sensor) p Find dead & noisy channels p Determine initial operating voltages (from pulse height plateau, I leak -V curve) Burn-in Test: Long-term (72 hr, 30’ between runs) test of whole ladder/wedge (conditions close to those in experiment) x-y movable laser head

24. 4 August 2000NIKHEF Overall Quality (first half-cylinder) Detector classification: p Dead channel: laser response < 40 ADC counts p Noisy channel: pedestal width > 6 ADC counts (normally < 2 counts excluding coherent noise) p Grade A: less than 2.6% dead/noisy channels p Grade B: less than 5.2% dead/noisy channels Use only detector grades A,B; mechanically OK Example for 9-chip detectors: DeadNoisy (better for other detector types)

25. 4 August 2000NIKHEF Production Status and Projection p Projected rates : assumed yieldcapacity  9-chip: 9.0/week 80% 15/wk  6-chip: 5.4/week 85% 10/wk  H-wedge: 6.2 week 85% 10/wk  F-wedge: 4.3/week 90% 15/wk Nov 1, 00 Aug 10, 00 July 7 50% line Rates include production, but in general dominated by sensor delivery. However, HDI “stuffing” at Promex (9-chip) also a concern (wire bond pull strength, HDI bubbling during surface mount) (as of July 7)

26. 4 August 2000NIKHEF Barrel Assembly in steps 1. Insert individual ladders into rotating fixture using 3D movable table 2. Manually push notches against posts (all under CMM) p Rule of thumb:  Align to 20  m (trigger)  Survey to 5  m (offline) p Precisely machined bulkheads p Barrel assembly done inside out (protect wire bonds)

27. 4 August 2000NIKHEF Barrel Assembly p Layer 4 glued to bulkheads (providing structural stiffness, holding passive BH) p Thermally conductive grease applied (active BH only) for other layers First 3 barrels assembled (  4 weeks/barrel, excluding survey) 3. Secure ladder using tapered pins

28. 4 August 2000NIKHEF Barrel Alignment p Shift d across ladder (3  m) p Shift  in radius (from ladder flatness, better than 60  m) p Rotation  in ladder plane (10  m  3  m) p Rotation  about short ladder axis (70  m  4.6  m) p Rotation  about long ladder axis (80  m  3.2  m)     Results for first barrel (similar for other two): Should be OK for trigger purposes Example: distribution for  : Note: relevant quantities are distributions’ RMS values (trigger accounts for average offsets)

29. 4 August 2000NIKHEF F-Disk Assembly p F-disk assembly less critical (not included in trigger), nevertheless performed under CMM p Quick process p After assembly, “central” F-disk cooling rings screwed onto active barrel bulkheads z=0 V depl HHHHHHL M L L L M Distribution of different quality devices over disks: H/M = Micron high/medium V depl, L = Eurisys low V depl

30. 4 August 2000NIKHEF Support Cylinder p Double-walled carbon fibre structure supporting all but H disks (supported by CFT layer 3) p Split support (cut at z=0) introduced very late: gain 6 months of “schedule time” (installation can be done with end-cap calorimeter cryostats on platform) p First half-cylinder ready

31. 4 August 2000NIKHEF Readout Electronics p For 5% occupancy, 1 kHz trigger rate: 10 10 bits/s  need error rate  10 -15 p Exercise readout system as much as possible before installation in experiment  10% system test using full readout chain (readout full F disk, barrel, barrel-disk assembly, H disk) p Complete readout chain (including L3 analysis, data storage) tested on several detectors Monitoring Control platform SEQSEQ SEQSEQ SEQSEQ SEQSEQ SEQSEQ SEQSEQ VRB Controller Optical Link 1Gb /s VBDVBD V R B 68k Secondary Datapath VME 3M 15531553 NRZ/ CLK IB L3HOST Examine HDI Low Mass

32. 4 August 2000NIKHEF In preparation Done Started Done Started Half doneSchedule

33. 4 August 2000NIKHEFCost