402.02.03.05 Module Assembly Extrapolating Experience from Phase 1 - PowerPoint PPT Presentation

402.02.03.05 Module Assembly – Extrapolating Experience from Phase 1 Upgrade Matthew Jones, Purdue University February 2-3, 2016 February ¡2-­‑3, ¡2016 ¡ M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ 1 ¡

Outline § Phase 1 Module Overview § Module Assembly Process § Lessons from Pre-production Module Assembly § Conceptual Phase 2 Module Construction § Suggested Design Constraints § Summary M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 2 ¡

Phase 1 Module Overview Module ¡end ¡holder ¡ Flat ¡flex ¡cable, ¡75 ¡cm ¡length ¡ (with ¡thru ¡hole ¡for ¡ (short ¡“connector ¡saver” ¡pigtail ¡flex ¡will ¡ #00 ¡or ¡M1.2 ¡screw) ¡ HDI ¡ TBM ¡ be ¡used ¡for ¡tesOng ¡during ¡assembly) ¡ FPIX ¡sensor ¡ ZIF ¡Connector ¡ 2x8 ¡ROCs ¡ Module ¡address ¡seYng ¡ HDI HDI Wirebond Sensor Sensor ROC Bump-bonds ROC Bump-bonds M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 3 ¡

Phase 1 Module Assembly Sensor § Sensors produced and tested by Sintef ROC § PSI46dig ROC’s fabricated in 250 Bump-bonds nm by IBM § Bump bonding performed by external vendor (RTI) § UBM deposition, solder bumping § ROC thinning § Flip-chip assembly and Pb-Sn reflow § Process developed over 10 years ago, used reliably since then § Bump bonded modules shipped directly to assembly sites M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 4 ¡

Phase 1 Module Assembly § High Density Interconnect § PCB manufactured by outside vendor § Design qualified to withstand bias voltage § Vendor performs electrical tests § Additional inspection and testing performed when received o Pad and trace widths o Metal thickness, surface quality o Wire bondability § Assembly of components (Fermilab) o Surface mount components o Placement and wire bonding of TBM o Assembly of mechanical interfaces (end holders) o Electrical tests M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 5 ¡

Phase 1 Module Assembly § Module inspection under microscope § Last chance to observe and document anomalies before assembly § Re-inspection of HDI ? ¡ § Check metal on pads § Check for non-flatness WL_BB_044 ¡ § Re-measure bare module IV § IV measured again after assembly § Needed to qualify assembly process M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 6 ¡

Phase 1 Module Assembly HDI HDI Wirebond Sensor Sensor ROC Bump-bonds ROC Bump-bonds § Assembly process: § Deposit epoxy on sensor (Araldite 2011) § Precisely position HDI on sensor, hold in place while epoxy cures § Form wire bonds between ROC’s and HDI § Encapsulate wire bonds § Thermal cycles, electrical tests § Two construction sites: Purdue and Nebraska § Substantially similar assembly process at both sites M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 7 ¡

Phase 1 Module Assembly § Automatic robotic assembly under manual supervision § Simplifies training, reduces reliance on highly skilled technicians § Well suited for a small pool of undergraduate labor (3-4 students) M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 8 ¡

Phase 1 Module Assembly § All assembly work performed on robotic gantry § Aerotech AGS10000 gantry, <10 µ m precision § Custom tooling holds parts using vacuum during assembly: bridge ¡ glue ¡ reservoir ¡ M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 9 ¡

Phase 1 Module Assembly § Alignment of pads on ROC’s and HDI for wire bonding § Semi-automated assembly process: § Position sensor module, ROC side down on one fixture § Position HDI, component side up, on a second fixture § Acquire fiducials on sensor module and HDI § Mix Araldite 2011 epoxy and spread in glue reservoir § Pick up stamp tool, coat with epoxy from reservoir § Stamp epoxy onto sensor side of bare module M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 10 ¡

Phase 1 Module Assembly § Semi-automated assembly process: § Pick up HDI using bridge tool and place on sensor § Transfer vacuum to bridge, rubber gasket applies force while epoxy cures (12 hours at room temperature) § Post-curing alignment check <50 µ m § Glued module placed on module carrier, where it will remain. M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 11 ¡

Phase 1 Module Assembly § Semi-automated process HDI Wirebond § F&K Delvotek 6400 wire bonder Sensor § Pattern recognition software determines bond pad locations ROC Bump-bonds § Automatic bonding using predetermined parameters § Bond quality monitored by online graphs of bond deformation vs time § Bond ROCs 0..7, then flip module carrier to bond ROCs 8..15. § Less than one hour per module when all goes as planned. M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 12 ¡

Phase 1 Module Assembly § Modules are tested before wire bonds are encapsulated. § Flex cable attached to ZIF connector – remains attached for all subsequent tests. § Tests based on PSI digital test board and associated software. § Test results are recorded in a database § Mainly looking for problems that can be fixed before encapsulation. § Such problems turn out to be rather uncommon. M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 13 ¡

Phase 1 Module Assembly § Wire bonds are encapsulated for environmental and mechanical protection § Sylgard 186 silicone based elastomer o Transparent and relative re-workable § Encapsulating fixtures hold module carriers – separate work area from assembly fixtures. § Dispense onto bond feet only using robotic gantry and dispensing syringe + pressure multiplier. o Acquire fiducials, measure needle tip height using optics o Dispense lines of encapsulant onto ends of bond feet on both ROC and HDI sides o Volume of encapsulant and alignment is critical: avoid wicking into region between sensor and ROC § Also encapsulate wire bonds on TBM § Cures in an oven at 50 deg. C for 2-3 hours M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 14 ¡

Phase 1 Module Assembly § Thermally cycle unpowered modules after encapsulation § Induces mechanical stress § 10 cycles from -30 to +50 C with 10 minute dwell times at each extreme § Chamber is purged with dry air throughout cycles § Primarily useful for qualifying the design – not necessary for production. § Time schedule § Can cycle 10+ modules at once § Cycles take ~12 hours M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 15 ¡

Phase 1 Module Assembly § Modules are shipped to other sites for testing and integration § X-ray testing at UIC and KU § Further testing, grading, and integration at Fermilab § Packaging § Modules encased in heat-sealed, static dissipative polyethylene tubing with a silica gel desiccant § Sealed modules placed into foam- lined box for transport § Use FedEx ground to ship and track modules M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 16 ¡

Phase 1 Module Assembly Staffing § Cost of operating a production site is dominated by staffing requirements § Technical staff needed for module production: § 1 senior engineer/technician § 2 wire bonder technicians (part time/hourly) § 2-4 undergraduates (hourly) § One postdoc (base funding) § One graduate student (base funding) § The robotic assembly process eliminates the need for multiple positions filled by highly skilled technical staff. § Unlike the model used for construction of the first FPix detector § Delays in delivery of parts led to frequent underutilization of staff M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 17 ¡

Lessons Learned from Phase 1 Production § Examples of problems: § Sensor bias voltage in close proximity to ground vias o Add additional layer of polyimide to HDI § Lack of representative parts with which to qualify wire bonding parameters o Partly due to unanticipated variability in quality of HDI’s o Difficulties wire-bonding pre-production modules § Several critical tolerances needed for reliable assembly o End-holder assembly procedure impacts the ability to pick up the HDI in later steps o Over-thinned ROC’s change height of bare module on assembly fixture – necessary to add custom shims to correct o Minor changes in component placement required changes in tooling o Epoxy coverage under wire bond pads critical for reliable wire bonding § General issues: § The design did not cleanly factor into independent, unrelated steps. § Future module designs may benefit from conscious effort to partition critical steps … § Desireable, but this may not always be possible. M. ¡Jones ¡-­‑ ¡402.02.03.05 ¡ February ¡2-­‑3, ¡2016 ¡ 18 ¡