Documents
Resources
Learning Center
Upload
Plans & pricing Sign in
Sign Out

Trip report Leb Vertex Fermilab Rad Hard Vertex

VIEWS: 3 PAGES: 25

									LEB and Vertex 2001
            Trip report


             Gabriele Chiodini
Fermilab, October 15, 2001 – Monday pixel meeting
         Attended conferences
7th Workshop on Electronics for              10th International workshop on
LHC Experiments. Stockholm,                  vertex detectors. Brunnen,
Sweden, 10-14 Sep. 2001.                     Switzerland 23-28 Sep. 2001.
Plenary sessions:
• Partially depleted SOI;                    Only plenary sessions:
• Trends: DSM, FPGA, and PCB.                • Operating vertex
• LHC status; electronics for                • Planned vertex
   Cal, Pix and AMS in space.                • Sensors
• Hera-B commissioning.                      • Detector readout
Parallel sessions:                           • Triggering with vertex
•   Tracker FE –Trigger electronics          • Hybridization and monolithic
•   B+Rad effs –Cal electronics
•   Optoelec. -DAQ and DCS                   • Other applications
•   Muons FE -GND+cooling+align.
How to win a Nobel price!!!




                         G.Chiodini - Oct 15, 2001 - Fermilab                 2
                      LEB01:
  Partially depleted SOI (K. Bernstein - IBM)

Bulk MOSFET                         •   QM limitation to scaling law
                                        requires new concepts : PD
                                        SOI , DTMOS, FinFET, …
                                        -------------------------------
                                    •   Floating body effect in SOI.
                                    •   PD SOI operation mode:
                                        equilibrium, dynamic, steady.
                                    •   Delay: spatial + temporal.
PD SOI                              •   Advantages:
                                          – Less C (eSIMOX = 4.1, eSi = 12).
                                          – Lower threshold.
                                          – No Latch-up.
                                    •   Applications: SRAM, powerPC,
                                        …
                                    •   Very few analog circuit in
                                        SOI.

                G.Chiodini - Oct 15, 2001 - Fermilab                           3
                    Trend in industry(1)
•   Scaling in DSM (K. Bernstein-IBM):
     – Now tox= 0.1 um (0.25 um is 4 generation beyond ).
     – Scaling limitation due to QM (Roll-off from Morthy’s law if not innovation).
     – Problems: Process spread (yield), power and temperature distribution,
       electromigration, hot carriers, SEU from surrounding material, oxide scaling
       (5 atoms, Igate is an issue), interconnect capacitive coupling, V drop.
     – Innovation:
         • Device : DP SOI, FinFeT, Strained Si MOSFET.
         • Package: Chip Stack Module, 3D neural network.
         • System: Compiler and multiprocessor parallelism (it’s the future:no monster chip).
• PC board (J.Bovier- Creative Electronic System)
     – Strong demand from new chip and technology: BGA and FBGA, fast dynamic,
       multi PS, good cooling, low EMI.
     – New board require a high level designs (no change after).
     – Software testability (for example: JTAG chain).
     – BGA and FPGA package limits the re-workability.
     – Follow the design rules and stay in contact with your provider.
     – 18 month PC board delivering for CES.

                                G.Chiodini - Oct 15, 2001 - Fermilab                 4
•
                    Trend in industry(2)
    Progress in FPGA (P.Alfke - Xiling):
     – PS: 1.5V (3.3V is dead, future 0.8V), good decoupling at LV is a problem.
     – Clock: 200MHz OK, > 500MHz need send clock+data (c is finite).
     – What we need:
          •   I/O levels.
          •   serial Gbyte I/O.
          •   embedded processor.
          •   RAM or easy external RAM interface.
     – Virtex-E (mature). Virtex-II (adolescent): multistandard I/O and efficient
       arithmetic. Virtex-II Pro (embryonic): MICROBLAZE 125 MHz on VirtexII.
•   SEU in commercial FPGA (for space, S. Mattsson):
     –   SETransient (for ex. glitch in the data or clock line) not seen in static mode.
     –   Ions, protons,m and neutrons irradiation.
     –   Antifuse an SRAM (often hang) based FPGA.
     –   Xiling, Lucent, ATMEL, LATTICE, ACTEL, MEC.
     –   Lower V more Upset, gold is bad, Xiling is on epilayer do not latch-up.
     –   Xiling: regular FF and Triple red. FF, readback (bitstream repair).
     –   Upset depend on frequency (non tested).
     –   Ion: Xiling lower Eth than ACTEL but 100 time smaller cross section plateau.
     –   Proton: Xiling cross section = 10E-14cm2, ACTEL 10 times more.
     –   Neutrons: ? No beam direction? No packaging effect?

                                    G.Chiodini - Oct 15, 2001 - Fermilab                   5
                LHC Status Reports(1)
• LHC (T.Taylor):
   – Commissioning: 1 January 2006.
   – LEP 90% dismounted.
   – Low beta 0.5m (0.25m you get SLHC, it’s feasible but BCO<25ns difficult).
   – Lattice dipole magnets from 3 companies: 3 working out of 4 (understood
     why), more expensive than expected. Test all first 300-400 than decide.
     Production slope not bad but if problem show up they go out of schedule.
   – Low beta quadrupoles from KEK and FNAL OK.
   – String test good. String2 test necessary for final installation.
• LHC Experiments: (J. Engelen)
   – Starting construction and learning how to do mass production.
   – FE more difficult than though (SLHC need to change all detectors).
   – CMS (4 reviews/year):
       • Surface hall ready but cavern delayed by natural rock substitute by concrete
       • Yokes ready. Coil ready for 2004.
       • 223 m2 Si tracking: 200 detectors to exercise production + APV25 in DSM good.
       • Pixel readout chip translated from DMILL to DSM.
       • Ecal: very noisy FE for APD. Muon: Drift chamber OK, RPC need oils to work (big
         battle).
       • DAQ+Trigger: pilot project ready.
                             G.Chiodini - Oct 15, 2001 - Fermilab                6
– ATLAS:
            LHC Status Reports(2)
   • Solenoid ready. 9m Prototype coil installed (goal 25m). 8 rings needed.
   • Tracker: pixel (DMILL to DSM FE issue) installed independently 1 year after the
     tracker, SCT FE barrel OK, SCT FE end cup not OK.
   • Accordion Lar (no gaps) started electrodes production (more or less).
   • Tile Hcal very well.
   • Muons: stand-alone high precision big chambers (quality control for mass
     production crucial).
   • DAQ+trigger go very well.
– LHCb:
   • Vertex: 300 um n-on-n single side, double metal layer microstrip.
   • Traker: silicon + straws.
   • Rich: HPD’s 1025 channels (500um x 500um), S/N=50, 25 MHz clock will be
     reached in next version.
   • Ecal and Hcal OK.
   • Muon: RPC(no so good so far) + MWPC.
– ALICE:
   • Starting construction, TDR and R&D to do.
   • A lot of technologies involved but in good shape : pixel, SSD, SDD, RICH (works
     at STAR), TPC, TOF ( multigap RPC is a novel concept dt=70ps), ECAL(CMS), L3
     magnet.

                          G.Chiodini - Oct 15, 2001 - Fermilab                7
–
          Electronics status reports
    Hybrid pixel (M. Campell):
    •   ATLAS, CMS, BTeV, ALICE/LHCb. BTeV is very relevant and must be watched.
    •   DSM: ALICE FE ready, LHCb next iteration BTeV FE almost ready.
    •   DMILL: conversion in DSM: ATLAS and CMS.
    •   3D pixel=wire chamber with junction. No dead area, likely rad-hard.
    •   Scaling: new current mode, dynamic range problems, LV problem.
– Cal (V. Radeka, “I encourage the people to be free to speak” ) :
    • Difficulty: large dynamic range (15-16 bit), uniformity and calibrations, fast.
    • CMS: 1fiber/crystal. No heat. Long tail killed by CMS (not for pile-up but because
      radiation dependent) but ALICE can effort the tail. FNAL chip tested but noise
      larger and next iteration needed (inside 800 Mbit/s serializer). How much FE
      inside and outside? Problem of power.
    • ATLAS: FE concentrated in the end of barrel (LHCb away few meter). In Tile-
      fiber the exponential tail is clipped (also if 5photoe/Mev!!!). Calibration problem
      with I constant and 0.1% accuracy due to parasitic L. Coherent noise is a problem,
      ground, shielding EMI to understand.
    • FE good again in the next interaction!!!. LV regulator no progress (nobody wants
      work on CERN solution). Discrete component better than IC. Optical link still
      problems. Availability of technology is an issue (Vddd=3V). Spares and money.
    • VLHC and SLHC: detectors are behind machine, rad effects, dynamic range issue.
    • O.1 um good for digital but analog and mix is hard: tunneling gives large Igate.
    • Suggestion: Faraday cage between FE and detector in big system (Cu layer is
      good).
                           G.Chiodini - Oct 15, 2001 - Fermilab                  8
  Hera-B electronic commissioning
– In almost every subsystem there were surprises after assemblies:
    •   Rare and random chips and passive components failure.
    •   Commercial optical link receiver: digital signals sensitive to particle flux.
    •   LV broke soon, spikes in 240V, sensing failed.
    •   Ground and shielding redesigned (also for crates).
    •   Cross talk analog and digital lines.
– Lessons:
    •   Hardware trigger must be tested also with no detector working.
    •   SMT connector better than ST connector.
    •   Quality of the board must be good.
    •   Online masking of hot channels.
    •   More expertise than one needed.
    •   Software for commissioning need a lot of work.
    •   GND, shielding and EMI critical.
    •   Hardware needs to support debugging + on-line monitoring.
    •   Use LVDS instead of open collector.
    •   Regular ¾ month reviews from other experiments.


                              G.Chiodini - Oct 15, 2001 - Fermilab                      9
                       Tracker FE (1)
– ALICE (all DSM)
   • SSD
       – HAL25 from ALICE28C 128 channels FE double side. JTAG programming and
         controller, Reg with majority vote logic for SEU (new), OK but some yield
         problems.
       – Ladder endcap electronics: controls 28 hybrids (6 FE chip/hybrid), power
         control, power protection (prevents latch-up, 1/m in PHOBE), readout, I/O
         buffering 25 m, JTAG. Solution 2 ASIC: analog buffer (now instable) +
         remaining functions (OK).
   • Pixel
       – Very nice test stand: pci card in pc, vme interface, jtag, daq adapter board.
         Used for wafer, bench, xray, Vpulse, beam. 65 kHz trigger rate. Labview +
         ROOT analysis. Calibration few minute per chip.
       – Pilot chip to controls 5+5chips on ladder 1 and 2. 800 Gbit/s G-link
         compatible serializer and optical link receiver.
   • SDD
       – large dynamic range FE and analog sampling (PASCAL chip OK). Huge amount
         of data after digitations (AMBRA chip, good up to 60 Krad but memory die at
         5 rad, now RAM in DSM). Two chips to avoid cross-talk.

                          G.Chiodini - Oct 15, 2001 - Fermilab                 10
                     Tracker FE (2)
– LHCb (all DSM)
   • BEETLE1.1 chip for SVD, VETO?, RICH and more tested fine. 128
     channel in pipeline, analog and binary readout not sparsified. 25 nsec
     peaking time (2 time worse than exp.). Submitted BEETLE-FE1.1-FE1.2-
     SR1.0 for improved peaking time and SEU hard regs and readback.
   • ALICE1-LHCb for pixel and RICH. Qthmin<1000e@Qnoise<129e with
     sensor (no 3 bit adjust). LV=1.8V (1.6V designed). Reg redundancy. Sigma
     upset 60 Mev proton = 3e10-16cm2, Eth=5MeV mm2/mg. 12 Mrad OK.
     SPS 150GeV pion beam, good efficiency, working frequency = 10 MHz.
– ATLAS
   • SCT: Single side 12cm long sensors.
       – ABCD3T chip in DMILL. Module tested in beam OK. Irradiation up to
         3E14/cm2 need 500V and the TW=10 ns, timing change with irradiation. Good
         for p, gamma, pions. Starting production.
       – SCT128A chip in DMILL. OK, 29%yield (3%FE+19%ADC+22%Dig+44%MUX).
   • Pixel: MCM chip in DMILL. LVDS output link. Works OK up to 90 MHZ. 8
     board irradiated at –30C up to 30 Mrad. After irradiation works up to 33
     MHz (goal 40 MHz, simulation 80 MHz). Yield <50 %, actually 11%.
                        G.Chiodini - Oct 15, 2001 - Fermilab             11
   Radiation and B effects on electronics
– N and low gamma irradiation on commercial amp, switch, Vreg, DAC and ADC
  (J.A. Agapito).
– Low dose rate effect in test structure (for ABDC chip of ATLAS)
    • Annealing is always beneficial and can not explain LDRE.
    • No LDRE observed in DMILL (max 0.5 krad/s up to 10 Mrad). Easy and more
      realistic irradiation test.
    • Damage higher if you don’t bias (h trapped drift due to E). Someone says that the
      opposite is true for CMOS.
– Antifuse-FPGA for CMS DT in u-det tracker (1 rad/Y). ACTEL A54SX32
  with redundancy architecture and partition. SEU due to low energy h and n
  negligible. SEU < 2.9E-12cm2/chip (2.2SEU/y) for Ep=60 MeV. TID up to 20
  krad OK, 40 krad still working.
– TID and SEU in FPGA in L1 muon for ATLAS (100rad/y). Xiling Virtex (2.5V)
  SRAM based:ASIC(1FF)=FPGA(30FF). SEU=4E-14cm2/bit and 1.6E-13
  cm2/chip stack for 60 MeV p. Readback and reprogramming. 40 MHz clock,
  SEU not frequency dependent (I don’t believe it). In not-hardened flash
  PROM test SEU<1.3E-18cm2 !!!! 50 krad gamma after annealing is OK.
– CAEN: HV, LV modules tested with p (60MeV, 7krad=2e10p/cm2), gamma, n
  (60 MeV, 2E1011n/cm2), and under B (10 kGauss 5%-10% loss in worse case).
  Good for CMS u-det but ATLAS has 10 times more n.
– Atlas muon TDC LSI (Toshiba Gate array 0.3um CMOS) instead of FPGA.
  Slow PLL, SEU measurable but tolerable, no latch-up.
                           G.Chiodini - Oct 15, 2001 - Fermilab               12
              Opto-electronics(1)
– 75um precision CMS u-barrel alignment: n irradiation p(18MeV)+Be
  up to 2.6E12 – 8E13 n/cm2 of COTS: LED (light decreases), LED
  driver, controller (SEU), lens, video sensor CMOS (synchronization
  and imaging lost at very end, OK with p but a lot of SEU) .
– GOL= Gbit/s data transmitter for LHC
                                   (P.Moreira: http://cern.ch/proj-gol).
   • DSM, 32 bit/I, G-link and Gbit-Ethernet, 0.8-1.6Gbit/s. Redundancy in
     logic not in the serialiser, wire wired bits.
   • Error free 4 day test but at 1.6MHz eye pattern not ideal (next version
     fix that).
   • 10 Mrad 10 keV gamma no change.
   • SEU < 3.2E-13cm2 at 800 MHz for 60 MeV p. Ion test: Eth measured (5-
     15MeVcm2/mg), increases with f, loss of lock is the dominant SEU.
   • SEU OK for serializer 10 times higher for PLL.
   • 100 samples to give to users. Few change for CMS and less SEU for PLL.

                       G.Chiodini - Oct 15, 2001 - Fermilab         13
               Opto-electronics(2)
– LHCb RICH optical link trasmitter: GOL(800Mbit/s) + VCSEL+multimode
  fibers. 1MHz not suppressed readout for L1.
– Laser driver array for optical transmission at LHC (G. Cervelli).
  DSM 3 channels for analogue and digital signals. Work very well, robust to
  Upset and total dose up to 20 Mrads. PDF of the production up to +/-3
  sigma measured.
– QA manual for the CMS tracker optical links for environmental tests in pre-
  production and production (cal, B, rad, T, ageing, …). QA used in prototyping
  phase (after irr. discovered laser and photodiode dead and glass dark).
– Analogue opto-hybrids for CMS tracker extensively tested now starts pre-
  production. 17000 assembled pieces from 2002 to 2004.
– ATLAS SCT optical link. 30Mev,24GeV p and gamma: Truelight VCSEL <<1%
  failure after 10y. 10mA thr instead of 5 mA. Redundancy connections not
  increase noise. BER good. Package of 6hybrid+6flex. Ready for mass
  production.
– DMILL R/T optical link for ATLAS pixel: 80 MHz data Tr., 40 MHz clock
  and Bi-Phase Mark decoder. 25 GeV p irradiation. He didn’t show up.
– Optical link for ATLAS Tile-cal alternative to FPGA Chicago solution. Series
  of problems, now try with CERN device like GOL.
                         G.Chiodini - Oct 15, 2001 - Fermilab          14
               Power distribution
– Power supplies CMS-ECAL APD-HV. Tested CAEN (SYS1527+HV4ch)
  and ISEG (HP-3858A HV8ch). Good ripple for both. Stability test
  CAEN is better (less variation, all channels behave the same).
– PS and distribution for ATLAS SCT.
   • 4088 module each one is own power cable.
   • 3 different cables: LM flex (Al)+LM special cable(Al, copper) +cable.
     http:/merlot.ijs.si/~cindro/low_mass.html.
   • HV and LV together and twisted (cost? heat?too dense pack?).
   • Too many connectors and cables. Too late to optimize, next prototype
     (Feb02) than pre-production (Apr02).
   • Maximum voltage drop highly above safety margin for ASIC’s. Rad Hard
     regulator located in PP2 in R&D state.




                       G.Chiodini - Oct 15, 2001 - Fermilab        15
  Operating vertex detectors (1)
– CDF
   • L00 pick-up noise from C fiber, removable by software but bad: S/N about 10.
   • 5 DSSD layers (lessons):
        – Each layer is unique (5 sensor, 6/10 hybrids): reduce parts.
        – Difficult assembly (25% of all module need repair work): work in pipe-line and use single
          side sensors.
   • ISL: Every was very carefully tested before but in the final installation the
     cooling fluid doesn’t circulate. Glue blockage seen in Al elbow with new boroscope.
     October fixed with laser.
   • Problem after installation:
        –   1.4% SVXII=1 wedge: damaged wires bonding on the port card
        –   2.8% SVXII=2 wedge: broken cable
        –   0.3%SVXII=½ ladder : has dead FE
        –   Cooling ISL central part
        –   Matching Inp/Out optical fiber impedance (introduce air or mylar, V up)
        –    PS communication unreliable
        –   Beam slope=0.6mrad (must be < 0.1mrad for vertexing); Beam center=3mm (not < 1mm,
            lifetime compromised). Move all detector by 3 mm.
   • Lessons:
        – Avoid custom made components
        – Long installation: delay from vendor too
        – Single component failures better affect efficiency and not acceptance.
                             G.Chiodini - Oct 15, 2001 - Fermilab                        16
 Operating vertex detectors (2)
– D0
  • 4Hdisk+6Barrel+12Fdisk DSSD. IP=30um up to eta<3.
  • Failure modes:
       – 30% yield p-stop isolation defects in 90 degrees stereo angle.
       – Micro discharging effects: p-implants edge not aligned with electrodes.
  • Lifetime limited by bypass capacitance.
  • Detector assembled and powered on:
       – 15% chips not in the readout because they can not be downloaded (cables?,
         connectors?, chips?…). Should recover more 50% in October.
       – Correlated noise. No sparsification up to now.
  • Very good mechanical alignment (thanks Sidet)
  • Lessons:
       – Use SSD
       – Much simpler design (with 6 sensor types were logistically very hard. Big
         enterprise!!!)




                         G.Chiodini - Oct 15, 2001 - Fermilab                17
  Operating vertex detectors (3)
– Babar
   •   Rad hard up to 2 Mrad
   •   ½ module not working out of 208. 1 dead module during operation
   •   Power failure + magnet quenching.
   •   Radiation:
         – No radiation damage, hit efficiency 98%, dE/dx 2sigma sep. pion/K.
         – Radiation monitoring with pin diode. NIEL scaling works (dI~fluence). Type
           inversion with e- (unexpected, no explanation).
         – Modules replacement 2004.
   • Alignment crucial for sin(2beta):
         – Minimize residual now one plane each time but in the future inversion of full
           matrix 6x(plane)E2 matrix.
         – Overlap hit very important for alignment, statistics not a problem.
         – Sin(2beta) measurements do not depend on the resolution function
           parameterization.
         – New method necessary to extract lifetime better than PDG.




                           G.Chiodini - Oct 15, 2001 - Fermilab                 18
  Operating vertex detectors (4)
– Belle
   • DSSD (yield in each layer: 98.8%, 96.3%, 93.5%).
   • FE: VA1 (IDEAS, Norway) work well but loose gain 10 time faster than
     expected (X ray beam losses was not noticed: orbit fixed + gold foil).
     Damaged modules replaced quickly and we could measure sin(2beta).
   • 2002 upgraded vertex: more rad-hard DSSD, more rad tolerant chip
     VA2, new beam pipe (R=1.5 cm) for less background. (Luminosity
     increases fast).
   • Difficult to predict any problem: radiation monitoring necessary (pin
     diode + radFET, do not use vertex).
– CleoIII
   • CVD diamond support. Reached < 2% Xo.
   • DSSD Hamamatzu:
          – FE: BE chip Honeywell Rad-hard.
          – RC chip (CSEM): AC and Rbias on separate chip
          – Encapsulated wire bonds.
   • 60% efficiency r-phi layer1 (z layers OK, layers 2-3 OK):
          – 3 broken Hybrid during installation.
          – 0% efficiency along rings getting worse with time (now also layer 2).
            Radiation sickness already at 10-20 kRad. Vbias limited at 100 V.
            Encapsulation do not permit reworking.
                            G.Chiodini - Oct 15, 2001 - Fermilab                    19
  Operating vertex detectors (5)
– Hera-B
   • Large Roman pot system not interfering with Hera storage ring.
   • Two iteration needed to arrive to the final system.
   • Redundant pumping system: turbos + penning + Ti sublimation (problems
     with Al cup out gassing at the begining).
   • Match beam impedance (shunt impedance Rs>100k for 52Mhz/f): Al tube
     with holes, 4 Steel strip, 8 BeCu wires (all equivalent Rs reduced a factor
     of 100). (MAFIA32 3D software).
   • No beam pick-up in the electronics (shielding works) but RF skin depth
     still not optimal (some RF resonances present).
   • Binary Ice system with two cooling and pumping circuits for redundancy:
       – Not recommended for T<-4C: viscosity increase, no benefits from binary
         system, safety hazard due to now explosive water-alcohol mixture.
– I will skip: H1, Zeus, STAR, PHOBOS and ATHENA.
   • First blast at RICH very scaring (PHOBOS: few % channels lost and gain
     changed).



                         G.Chiodini - Oct 15, 2001 - Fermilab              20
–
                 Planned vertex (1)
    CDF and D0 upgrade:
    •   SSD and high voltage detetcors
    •   SVX4 in DSM
    •   Few hybrid types.
    •   Less material (for ex.: port card out of tracking region)
    •   No double metal but kapton. No optical link but Cu transmission line (flex circuit).
– ATLAS SCT:
    • Sensors: Hamamatzu, CiS, Sintef
    • Termocycle for every hybrid and for subset of modules.
    • Pre-production is starting.
– ATLAS pixel:
    • Insertable layout. Less module more material in the forward region. One year
      later others.
    • Al cooling pipe instead of C (broke at 8 Atm). Fluid C3F4.
    • Going in DSM (right now Qthrms=1000e- after 5 bit trimming) 50 Mrad radiation
      test OK.
    • Module similar to BTeV. Thermal: C is strong and can break wire bonds (no so far).
    • Chip in DSM: FE, MCC, VDC, DORIC. Engineering run in few weeks.
    • Sensor tiles: CiS gave good sensors (we don’t know the manufacture yield, maybe
      3 out of 15), Tesla gave no one (now maybe understood the problem).
    • 50 dummy module from two vendors. Assembly are going on but very slow process.
    • Thinning not a problems (California vendors): plasma or lapping.
                             G.Chiodini - Oct 15, 2001 - Fermilab                  21
                Planned vertex (2)
– CMS silicon tracker:
   • A lot of tests: SSD Hamamatzu with p/n, AC coupled, polysilicon R , 300-
     500um, low resistivity, rho=1.5-3KOhmcm, metal overhang the p+ implant
     (500 to 600 V Vbreak).
   • Crystal orientation <100> instead of usual <111)
       – Radiation up to 3E14 1 MeV eq n/cm2, Vdepl=280V.
       – Less interstrip capacitance.
   • Layout details in the talk.
   • APV25 chip in DSM has pion SEU cross section few times larger than us.
– CMS pixel:
   • SEU handled by frequent chip reset.
   • Sensor: 11GR single open p-ring hold 1000V (double break at 300 V)
       – Single ring break at > 600 V after 6E1014p/cm2
       – Double and double crossed have 2 soft and hard breakdown after irradiation
         (not present in diode, maybe premature breakdown for some pixels).
   • Bump bonding in house capability (PSI)
   • P-spray is going to be developed at PSI.

                         G.Chiodini - Oct 15, 2001 - Fermilab              22
                    Planned vertex (3)
– ALICE pixel:
    •   Qth<1000+/-200e- after 3 bit trimming, 32x256 pixels.
    •   6 wafer(86 chip each) tested at 300um and 750um thickness.
    •   Yield: class I(30%), II(20 no injection more than 6000ch) ,III(20%), IV(30%bad).
    •   Assemblies with AMS and VTT of p/n Canberra sensors (150-200um thick) and
        chips on PC board acceptable (Thr and noise similar to bare chips). Test beam OK.
– LHCb tracker:
    • 9 stations, all different.
    • SSD p/n 6’’ wafer. –5C cold hydrogen box (Al foil + foam material).
    • Test beam with SPA (Kiev) detectors oxygenated and with multiguard rings. They
      have too early breakdown.
– LHCb vertex locator (impact parameter trigger):
    •     1.3E13neq/cm2/y at 0.8 cm.
    •     phi and r sensors. The r as 9 mm strip near the center.
    •    Test beam: Double side Delphi, N+/n Hamamatzu and p/n micron.
    •    SSD n+/n (from test beam), double metal and analog readout.
    •    Impact parameter trigger with 150-250 pc. B would help a lot. New idea: superL1.
    •    L0 pile up VETO efficiency of 30-40% of single b-bbar events at optimal
        luminosity.

                            G.Chiodini - Oct 15, 2001 - Fermilab                23
– ROSE
                              Miscellanea
   • Carbonated Si is less rad tolerant, Oxigenated more (large reduction reverse
     annealing, O’NEIL violation).
   • Large variation of standard silicon likely due to C concentration (Sintef always
     good). For same material more stable after oxygenation.
   • The ratio C/O rules the behavior.
   • Puzzle neutron-proton explained by the interchange-transfer model:
         – Coulomb interaction produces vacancies (sucked by O defects)
         – Nuclear interaction produce clusters
   • Oxygenation useful only for n+/n/p sensor and not p+/n (help only electron
     collection).
– CDF impact parameter and track trigger
   •   Almost commissioned after a lot of problems. No L2 processor yet.
   •   Not completely sure we pick up the right event.
   •   The material budget was more than expected.
   •   Lessons:
         –   Z alignment is challenge (beam slope correctable).
         –   Less number of boards and more firmware.
         –   Plan for commissioning and debugging, not only data taking.
         –   PLD/FPGA 6 years old.
         –   More software effort.
         –   Test with no real detector and away of the trigger room necessary.
         –   No switch on the board but programmable.
         –   Data driven good, clock enable good (no strobe data).

                               G.Chiodini - Oct 15, 2001 - Fermilab               24
       Summary talk vertex2001
– C. Damerell
   • Operating vertex very complex
       –   Difficult to access and long sensor chain in one ladder.
       –   Very sensitive to errors and beam losses.
       –   QA precision necessary.
       –   Group meeting, reviews, written document and worksheets necessary.
       –   SSD preferred, but if DSD floating electronics recommended.
       –   Wire bonds yield and thermal cycling necessary.
   • Near future
       – Vertex with pixel (except LHCb), tracking with microstrip and gas chambers.
       – Deep Submicron CMOS electronics. Concerns: LV and I protection.
       – Vertex triggering on-line: BTeV is going in the right direction.
   • Linear collider
       – Hybrid pixel are almost ready but too much material (20 times more).
       – High speed column based CCD and Monolithic Active Pixel Sensor are the
         preferred solution (not depleted device). Lot of progress but Rad-hardness
         must be tested (different from CMOS).
   • A lot of not high energy applications.

                          G.Chiodini - Oct 15, 2001 - Fermilab              25

								
To top