LHCb Silicon Tracker hybrids overview LHCb ST Detectors

LHCb Silicon Tracker hybrids overview LHCb ST Detectors: IT & TT hybrid design procurement & testing words on vendor issues conclusions Frank Lehner U Zurich LECC Heidelberg Sep 14, 2005

LHCb Experiment pp collision dedicated B physics experiment – single arm forward spectrometer vertex detector, RICH, tracking, calorimeter, muon four tracking stations: Trigger Tracker (TT) in front of magnet – solely silicon strips T 1 -T 3 station after magnet – inner part employs silicon strips

Silicon Tracker TT station up to 160 cm long silicon modules in four layers: 0, 5, -5, 0 degrees hybrids, cooling outside acceptance ~145 k channels IT stations of T 1 -T 3 covers innermost part after magnet cross shaped around beam pipe 2% of surface area but 20% tracks 10 or 20 cm long silicon modules ~130 k channels

Detector overview IT-detector module: one (10 cm) or two (20 cm) sensor ladders, 320 or 410 m thick silicon with pitch ~200 m hybrid at ladder end, supported on aluminum balcony TT-detector module: full length 160 cm 500 m thick silicon with pitch ~180 m two half-modules w/ 7 sensors each are mated to full length full module has at both ends hybrid stacks (out of detector acceptance) hybrids supported on carrier pieces further readout via 5 m long cables to service boxes (digitization, see talk by A. Vollhardt) Total electronic channels: ~280 k IT: 336 readout units w/ 3 chips each TT: 280 readout units w/ 4 chips each IT ladder TT half module

Beetle chip common development for LHCb silicon detectors designed by ASIC lab Heidelberg & NIKHEF to meet LHC requirements fabricated in 0. 25 m CMOS – radiation tested up to 40 MRad 128 channel low noise preamplifier w/ CR-RC shaper analogue pipeline, programmable latency up to 4 s L 0 trigger rate 1. 1 MHz multiplexed deadtimeless R/O within 900 ns input pads: four row staggering, effective pitch ~40 m size 5. 5 x 6. 1 mm 2

IT hybrid three Beetle chips on flex (polyimide) PCB: 130 x 77 mm 2 opening in bottom solder resist @ beetle positions for thermal and GND contact ceramic pitch adapter flexible pigtail connector single row 60 -pin, pitch 0. 5 mm PCB mounted on aluminum heat spreader 386 hybrids needed (incl. spares) ~15% as pre-series delivered

TT hybrid four Beetle chips on flex PCB: 67 x 70 mm 2 stack of up to three hybrids at both ends of a TT full module three variants/flavor with different size necessary: long/medium/short lowest level: aluminum nitride carrier + pitchadapter to silicon upper two levels: copper plate + pitch adapter to flexible interconnect robust dual row 80 pin connector w/ 1 mm pitch 322 hybrids needed (incl. spare) ~15% as pre-series delivered

Pitch adapters pitch adapters (PA) as fan-ins from ~200 m silicon strip pitch to 40 m narrow Beetle input pitch all three types of PAs for IT & TT: sizes: from 77 x 11 mm 2 to 94 x 24 mm 2 0. 25 mm Alumina in thin film metal: Ti. W (50 nm)/Au (2 m) Al wire bonding very satisfactory no severe degradation after accelerated aging fan-in structure is mirror image of fourfold staggering of Beetle Trace width/separation: less than 10 m in limited areas 30 -40 m trace width elsewhere Vendor: RHe Microsystems Gmb. H

Flexible PCBs four layers (from top: routing/GND/VDD) flex PCB polyimide 50 m, Cu 18 m sizes: 67 x 60 mm 2 (TT) up to 130 x 77 mm 2 (IT) Au/Ni surface for Al bonding min. trace width/separation 75 m/75 m solder resist on top (laquer), polyimide bottom layer only one active component (Beetle) passive components (all SMD): R’s, ceramic C’s, PT-1000, B 2 B connector design “core” successfully used since 1 st testbeam 2001 prototypes, pre-series and series PCBs successfully produced at Optiprint AG

Hybrid assembly all assembly steps done at single company (RHe Microsystems) procedures, tolerances, acceptance criteria defined in specs document manual chip placement, accurate ± 20 m aluminum wire bonding 25 m wire for chip – PA bonds 30 m wire for flex PCB - chip regular pull tests performed thermal cycling: 10 x from -20ºC to +60ºC, 5 minutes at each endpoint test stand set up at vendor test of basic digital and analog functionality of all beetles on hybrid repair (=chip replacement) still possible

Hybrid assembly full documentation of “which chip goes to what position on hybrid from wafer maps (-> hybrid database) excellent documentation of pull tests, electrical tests & repair work

Hybrid testing – preassembly tests: chips chip testing on uncut beetle wafer at MPI Heidelberg setup wafer prober & needle card & DAQ test of full functionality of analog and digital performance: power consumption I 2 C addressing tests control logic sequential pipeline scans pulse shape scans (…) tight quality acceptance criteria typical yield: ~90% 3 x more good chips than needed

Hybrid testing – preassembly tests: PCB & PA flex PCB tests at vendor (Optiprint AG) automatic optical inspection electrical test for contact only perfect PCBs accepted, no bad PCB found up to now (>100) specs (bond pad width, thickness etc. ) defined in specs document bond tests on lot basis (at hybrid assembly company) pitch adapters (RHe Microsystems Gmb. H) bond pull tests electrical test with needle card (on spare pads) complementary optical inspection several shorted traces allowed prior to repair laser repair of shorts – final spec: 1 short or interrupt 70% w/o defect at all; remaining have up to 3 repairs

Tests during hybrid assembly test stand set up at vendor (RHe Microsystems Gmb. H) short and important test of basic analog and digital functionality test performed before and after thermal cycling: 10 x cycles – 20 C <-> +60 C, 5 min. at each endpoint standard test is done after assembly but prior to chip-pitch adapter bonds easy detection of missing bonds also full chip replacement still possible/feasible at this stage so far no sign of “infant mortality” of 43 (43) thermally cycled TT-hybrids

Hybrid burn-in test extensive hybrid testing program after delivery to MPI Heidelberg characterize, document and grade hybrids start with visual inspection detailed R/O test at start/middle/end of 48 h long burn-in run at 1 MHz rate R/O tests similar to Beetle chip testing wafer tests in box at room temperature, no thermal cycles up to 16 hybrids (any combination of IT & TT hybrids) can run at same time analysis output linked into database

Hybrid burn-in test full digital and analog tests (similar to chip wafer tests) I 2 C addressing, power consumption etc. raw noise & baseline subtracted noise mapping of pipeline - relative pipeline cell gain: ~5% variation pulse shapes & timing characteristics (…) check power-on startup run at ~1 MHz trigger rate: check Beetle synchronicity check HV contacts; monitor HV at 600 V; check for sparking effects during burnin check PT-1000, VDDD sense lines yield: 84 out of 92 hybrids passed on first go -> analysis of failed hybrids not yet concluded

Problems IT hybrid has openings under Beetle chip positions in bottom solder resist filled with silver epoxy connection to common GND thermal conductive path to heat sink problems on some prototype and preseries hybrid: openings too large problems with insulation potential risk of GND/VDD short – repair with HV insulating cover fixed for series production

vendor issues I choice of right vendors is a big issue for small-scale projects with design features that are at cutting edge we experienced large difficulties in hybrid prototyping phase which caused lots of delays; fortunately, they were no real show stoppers of the project pursued initially approach of two different companies ceramic pitch adapter (PA) production in thin film technology hybrid assembly, SMD loading & bonding problems with coordinating quality standards and agreeing on common specs among companies like pull strength on PA etc.

vendor issues II vendor qualification process of first company (dedicated PA production) suffered significantly from low quality and large delivery delays of ceramics mainly due to overestimation of own abilities at company partially due to design/layout that could have been made simpler second (dedicated SMT) company lost interest in continuing after successful prototypes due to low quality ceramics/PA due to technical difficulties (4 x staggered chip-PA bonding) luckily, we could identify a single company capable of PA production, SMT and wire bonding in-house (RHe Microsystems Gmb. H) developed tight relationship customer - company from day one worked/cooperated together on several design modifications to simplify fabrication process rapid & successful prototypes on very short time scale proof of principle achieved, smooth transition to series production

vendor issue III our lessons: start with vendor qualification as early as possible be prepared for back up solutions work closely together with companies: what can they do? what is achievable at reasonable yield/cost? what do they recommend? what can be changed/modified in the design/process to make it simpler/fail safer? cultivate a tight connection & close relationship understand early if the company shows real interest in the project. This interest should go beyond commercial interests

Summary hybrid assembly outsourced to single company capable of pitch adapter production, SMT mounting and aluminum wire-bonding hybrid pre-series (15%) for IT & TT detectors has been delivered just recently thorough hybrid testing program is installed 84 out of 92 tested hybrids passed our testing on first go testing not yet fully concluded - remaining hybrids being debugged expect to give go-ahead for series production very soon