US CMS HL-LHC PROPOSAL Phase 2 Pixel/Strip Module Development ABSTRACT We propose to work on the development of the Pixel/Strip modules that will populate the inner part of the upgrade CMS tracker in upgrade Phase II. These modules utilize correlations between pairs of closely spaced silicon sensors to produce “stubs” with nominal Pt above 2-3 GeV. These stubs reduce the data rate for transmission to offdetector track finding by a factor of ~twenty and enable the track-based triggering which is a cornerstone of the Phase 2 upgrade. We also describe plans for expanded future work on 3D active edge sensors. 1 R&D OVERVIEW HL-LHC, with an average of 140 interactions per crossing, poses extraordinary experimental challenges. Pileup backgrounds will degrade calorimeter resolutions, triggers on missing Et will become more difficult, and raising thresholds on the muon trigger beyond 30 GeV no longer provides significant trigger rate reduction. Track information uniquely provides the ability to isolate the primary vertex and reduce or eliminate backgrounds from pileup. Z vertex resolution is crucial. Full track reconstruction with pixel hits can provide single vertex Z resolution. The PS module is designed to provide intermediate, ~1 mm, Z tracking resolution for the Level 1 trigger. This will reduce the combinatoric background by ~99% and also provide a region of interest for the pixel readout which can be used for local readout either at Level 1, if there is enough margin in the L1 latency, or to supplement Level 2. Figure 1. CMS Phase 2 upgrade tracker design. PS modules are blue. The planned CMS Phase 2 tracker is shown in figure 1. The PS modules occupy the region between 20 and 60 cm in radius. Identical modules will be used in the barrel and disk regions. A total of ~15,000 modules will be needed for the full tracker. These devices will have unprecedented functionality and complexity. PS modules are composed of long strip (LS) and short strip (SS) sensors. Hits from the long and short strip sensors are clustered, compared, and combined in the MPA readout chip to identify “stubs” where the offsets of the rphi hits in the two layers correspond to transverse momenta greater than 2-3 GeV. Stubs are processed through an in-chip pipeline to insure that on average a full event is read out every 25 ns. The z resolution of the short strip sensor provides the information needed to resolve the primary vertex when off-detector tracks are formed, but is not essential for stub formation. The chip must also store and transmit full event data for those events accepted by Level 1. The PS module presents significant electronic, mechanical, and interconnect challenges. The complexity of the chip itself is considerable. Connectivity determines the module size. Two columns of eight reticule (typically at most 2x3 cm) sized MPA chips define the ~5cm x 10 cm module area. The two layers of sensors must be separated by varying amounts, depending on the radius and orientation of the module. The spacers between modules must be low mass and be thermally conductive. A traditional pixel architecture that utilizes two columns of readout chips with connections only at the outer edge cannot transmit hit information between adjacent columns without the unacceptable complexity of routing that information around the periphery of the module. This causes a dead region for stubs where tracks pass between the low z and high z columns within a module. This can be solved by adding connectivity, either with through-silicon vias, an interposer, or utilizing a hybrid design using low profile wirebonds. In previous years we had developed a complete track trigger conceptual design within the context of the “Long Barrel” concept, with a “VICTR II” chip Verilog design, system simulation, 3D chip prototypes, support structures based on carbon fiber box beams, as well as a fairly complete conceptual design of off-detector tracking. We hope to complete testing of these prototypes as we redirect our efforts to the barrel-disk design. The barrel disk design poses different challenges and our involvement will be more circumscribed. However we expect to build on our previous work to develop interconnects, carbon foam technology, bump bonding and assembly, and integrated circuit designs. In FY14 we plan to concentrate on PS module and support design, prototyping, and testing. We will work closely with CERN to develop these designs for the TDR. We also hope to explore TSV technologies and possible interposer-based solutions. We plan to develop industrial connections within the US for module assembly, TSV fabrication, and interposer production. Finally, although not a component of this year’s proposal, we plan to continue the development of 4-side-buttable “active tiles” which combine active edge technology with 3D integration. These could lead to lower cost, fine pitch pixelated sensor arrays that could be of great interest to both the HEP and BES communities. Our plan is to pursue active edge tiles with generic R&D funds through FY14, but expression of interest, eventual financial support, and collaboration within the CMS context would be very valuable. 2 PARTICIPATING INSTITUTIONS AND PRINCIPAL INVESTIGATORS Cornell University – Jim Alexander, Anders Ryd, Julia Thom, Peter Wittich Brown University – Ulrich Heintz, Meena Naraian University of California, Davis – Max Chertok, Mani Tripathi University of California, Riverside – John Ellison, Gail Hanson University of California, Santa Barbara – Joe Incandela Fermilab – William Cooper, Stefan Gruenendahl, Marvin Johnson, Ron Lipton, Lenny Spiegel 3 TECHNICAL DESCRIPTION AND DELIVERABLES OF THE PROPOSED R&D A. Full scale dummy module The PS module is mechanically, thermally and electrically challenging. Signals from the long strip sensor are amplified, discriminated, and then routed down to the MPA short strip chips which provide the clustering, stub finding, and readout functionality. DC-DC converters, concentrator chips and GBT transceivers are located on the short edge of the modules. The most recent design simulations show acceptable thermal performance with either carbon foam or carbon-fiber-aluminum inter-sensor spacers. The short strip assembly consists of a 2x8 array of MPA chips bump bonded to a sensor with 1.5 x 0.1 mm pitch. We propose to build and test a full-scale dummy sensor/MPA assembly. This will be done in close collaboration with the DESY and CERN module and electronics design efforts. We will fabricate both dummy sensors and dummy MPA chips with the correct bump bond pitch and chip spacing. These parts will also have serpentine heating elements to simulate the power dissipation in the chips and sensors as well Figure 2 PS module design (version 2.4) shown on a carbon fiber-faced carbon foam cooling tube support. Figure 3 Dummy module sensor and chip layout showing staggered bump bonds, serpentine heating elements, and pads for daisy chain testing. as daisy chains for interconnect testing and regions where alternate interconnect ideas can be tested. These parts can be used as building blocks for thermal testing and development of assembly techniques for the full module. We will then work with industry to assemble prototype dummy sensor/MPA assemblies that will be used for thermal and mechanical studies. Deliverable – Dummy readout chips and bumped sensors Deliverable – Assembled dummy PS module Deliverable – Dummy module thermal, electrical, and mechanical tests B. MaPSA-Lite Development In parallel with the development of dummy module components the first step in the development of integrated devices will be the development of a small demonstrator Figure 4. MaPSA-Lite sensor (yellow) and ROIC (green) module with 4 readout chips with 16 phi and 3 z channels each. The chip will be designed at CERN in 65 nm with a full analog front end and simplified digital serial readout. In parallel with our work on the dummy module we will explore fabrication of the MaPSA-lite assembly with US companies as part of a feasibility study for full module assembly. The MaPSA-lite can be tested with the CERN GLIB or more flexibly with a National Instruments Flex-Rio system. If necessary we will develop circuit boards to interface the MaPSA-lite with the National Instruments test systems available at Fermilab, Brown, and Cornell. Deliverable – MaPSA-lite PCB Deliverable – MaPSA-lite assembly Deliverable – MaPSA-lite tests C. Mechanical and thermal testing The PS module depends on novel, low mass high thermal conductivity materials for acceptable thermal performance, especially after irradiation. Our previous design was based on carbon foam and we have built and tested dummy modules with this material. We have also procured and done some initial characterization of ALCF material. We propose a program of test devices and measurements to understand the thermal and mechanical properties of the carbon foam, how to incorporate carbon foams into support structures, and optimal techniques to fabricate high thermal conductivity interconnects. It will be desirable to have uniform thin layers of epoxy applied to silicon and carbon foam surfaces for the purpose of making stacks. The method needs to be repeatable and have a good rate of production for large-scale module fabrication. Earlier techniques, employed in HEP and elsewhere, have relied on shims to define the thickness – epoxy is applied to surfaces and excess is removed using a razor blade, or equivalent tool, that is wiped along the shims. We propose to use a robotic dispenser made by EFD Nordson, which is in use at the Facility for Interconnect Technologies at UC Davis. The dispenser uses a piezoelectric actuator to squeeze controlled amounts of paste from a syringe – we have consistently deposited 130 um solder balls using this system. In this study, we propose to first deposit a matrix of controlled amounts of epoxy and hardener on glass slides, observe the spread of the mixture after applying pressure, and measure the resultant thickness of the epoxy layer. The quantity deposited per squeeze and the pitch of the drops can be varied until desired thickness of about 20-30 um is achieved. We will next apply the technique to carbon foam attachments to silicon wafers, and perform mechanical strength measurements. Figure 5.Thermal conductivity test setup Figure 6 UC Davis clean room In order to fully implement the epoxy in the thermal modeling of the stacks, we need to carefully measure the coefficient of thermal conductivity of a thin layer of epoxy contacting to carbon foams and other materials. We are proposing to build a test set-up that will be used for various samples, as well as for testing fully assembled stacks. The set-up is based on earlier efforts of this nature, albeit with some modifications that simplify the apparatus. Figure 5 shows the basic elements of the test jig. The assembly is enclosed in a vacuum chamber to reduce heat loss due to convection. The heat flow is measured above and below the sensor to ensure thermal isolation. Temperatures are measured at 4 points each above and below the sample to get enough data points to fit the thermal gradient and extract the temperature drop across the sample. The coefficient of thermal conductivity can be extracted knowing the heat flow and temperature drop across the sample. We will design and build the system at UC Davis with help from our machine shop and electronics shop. A readout system using LabView will be implemented to control various sensors and record the measurements. We plan to design, fabricate, and test carbon fiber-faced carbon foam rod support structures with imbedded CO2 cooling pipes. These parts are in an early stage of conceptual design and will require engineering design, short prototype fabrication, and thermal and mechanical testing. Deliverable: Thermal and mechanical tests of thin epoxy carbon foam layers Deliverable: Design, fabrication, and testing of rod support structures D. Interconnect R&D Interposer with analog signals Long Strip er Long Strip er Pixel er ~mm Pixel er Pixel er Flex ~mm interconnect Carbon Foam TSV Long Strip er ~10 cm Long strip signals pass through flex Long strip are ½ module length ~10 cm ~10 cm Figure 7. Possible interconnect topologies a) edge interconnect b) TSV-based c) Interposer-based The PS module poses special interconnect challenges. Long strip information must be conveyed to the pixel chip that forms the inter-tier coincidence. Information on hits and clusters must be conveyed between chips in both r-phi and z to avoid holes in stub acceptance. This is particularly challenging in the z direction, where information is passed between columns. Figure 7a shows the “baseline” configuration, where all information is passed along the edge of the module. There is no provision for Z interconnect and all chip IO, strip input from the top tier, and power connections are concentrated on a single edge. Inserting TSVs as in figure 7b can solve the problem of z interconnect, distribute power and ground on the surface of the chip and relieve some of the congestion at the edge. An interposer-based design (figure 7c) provides Z interconnection, removal of congestion at the edge by routing the digital signals to the center of the module, and the possibility of integrating the functions of the long and short strip chips into one IC. We propose to pursue both the TSV and interposer approaches. We had started collaborating with CERN on their 3-T TSV development project last year. However the company that originally quoted on the TSV work raised their price from $85k to $150k, beyond our available funding. The current price for dedicated TSV insertion at Tezzaron/Novati is also about $150k, and therefore is probably too expensive to fund this year unless it is more clearly on the baseline development path. However there is now a prospect of fabricating TSVs in a multiproject run. This would considerably lower the cost and provide a clean development path for TSV–based chips. We would pursue the TSV work this year only if there is an opportunity to demonstrate the technology affordably. ROIC wirebonds ROIC Sensor Figure 8a Low profile wirebonds Figure 8b. Sketch of a wirebond-based column interconnect We have developed a conceptual design for an interposer-based module with options for a silicon or glass/TSV-based device as well as a more standard PCB interposer. Both of these designs are based on moving the p-phi connections outboard and ganging the inner pixel to allow space for digital connections in the center of the module. Space between chips in r-phi is generated by rerouting the bond pads slightly inboard using 96 rather than 100 micron pitch. The PCB approach has the disadvantage of larger via size (~80 microns), and larger feature size than silicon or glass. The thicker dielectrics in PCB mean that controlled impedance traces must be larger. Both effects make it more difficult to compress the digital section to a small (~6 mm) area. PCBs also have significant thermal expansion coefficient mismatch with silicon (10-20 vs 3 ppm/degree C) that can result in unacceptable warping of the assembly. Either silicon or glass-based interposers can solve these problems. Silicon interposers have a perfect CTE match, can provide <10 micron routing, 10-50 micron vias with thin (micron-level) dielectrics. Glass interposers have similar properties but with smaller dielectric constant, simpler via etching but poorer thermal conductivity. We plan to investigate glass or silicon interposer solutions as part of the dummy module work only if costs for prototypes are affordable (less than $30,000). This appears likely. We have recently had discussions with companies building glass interposers and may be able to share an interposer development test run with a run for x-ray imaging focal planes funded by BES/Argonne. The Facility for Interconnect Technology at UC Davis maintains capabilities for flipchip bump-bonding using a variety of techniques, including gold-stud bonding, indium-bump bonding, solder-ball reflow bonding, and anisotropic conductive film attachments. We use the Northern California Nanotechnology Center on UC Davis campus for mask making, photolithography and wet chemistry for wafer fabrication. We will design and fabricate dummy sensors and dummy read-out chips, which will be used to make mock-ups of the stacks. We will pattern traces on the dummy ROIC to be used as heaters to mimic the power dissipation. The dummy sensors will also have loop-back test structures to measure bump-bonding yield. As part of this proposal, we will finish the work on bump-bonding of VICTR chips with pcb interposers and sensors from BNL We have also started to study a more conventional approach, where low profile wirebonds nestle under the bump bonds and provide a signal path between the MPA chip columns. In our initial attempt (figure 8a) the wirebond height was measured to be between 50 and 93 microns. We then added a 50 micron spacer to the top chip and glued a glass slide to the surface, compressing the bonds to 50 microns. There was no apparent issue with bond breakage. This solution has its own issues. The proximity of the digital signal-carrying bonds to the sensor surface means that the sensor would have to be shielded, presumably by a cu-kapton ground plane glued to the sensor surface. The bump bonds should have a standoff of 100-150 microns. Such a design would have to be tested in a bump bonded assembly. Digital pickup would also have to be measured. The dummy module discussed in part A has provisions for such testing. Although not as elegant as the interposer and TSV solutions this (or a variant) may be the most inexpensive means of achieving intercolumn communication. Deliverable – Interposer or TSV demonstration devices Deliverable – Wirebond interconnect studies E. Testing of the VICTR 3D chip The VICTR chip was designed and fabricated as a demonstration of the use of 3-D integrated circuit technology for the CMS track trigger. Through-silicon-vias provide the ability for the circuitry to connect both to the top and bottom directly and locally, enabling local formation of stubs with no need for edge communication. The 3D interconnect therefore allows for fabrication of larger sized (10x10 cm) modules with direct vertical analog connections between the long and short strip tiers. This technology is no longer the baseline, due to the slow development of the technology and the issue of yield of the 25 chips that would have to be bonded to the sensor surface. This is addressed in active edge R&D. Figure 9. Test results for VICTR showing noise and threshold distributions Figure 10. Sensor mounted on top of VICR through a PC board interposer VICTR consists of two tiers of 0.13 micron CMOS circuitry. The two tiers are bonded face-to-face on 4 micron pitch with Direct Oxide Bonding (DBI) technology. We have completed the first phase of VICTR testing without sensor and have confirmed the basic functionality of the chip including communication between tiers. We are in the process of testing the long strip tier of the VICTR with a sensor bump bonded to the top through an interposer. In the final phase of this work the VICTR chips (and its cousins VIPIC and VIP) are being DBI bonded to a sensor wafer provided by Brookhaven. This work is in progress at Ziptronix and delivery is scheduled for January. We propose to test this assembly on the bench, and if the bench tests are successful add the interposer/top sensor assembly to complete the 5 layer 3-D stack. The resulting assemblies would then be tested in the Fermilab test beam. Deliverable – Test of the short strip – VICTR stack Deliverable – Assembly and test of the 5-layer stack Deliverable – Beam tests of stacked sensors and ROICs F. ASIC Design We are in discussion with the CERN microelectronics group on the development of a radiation hard static RAM in 65 nm. This “IP” would be utilized both in the MPA chip and a future pixel readout chip. The development of this custom SRAM requires custom design tools and memory “compilers”. The Fermilab ASIC group is investigating the resources necessary to develop the capability for such custom memory compilation. Assuming the answer is positive we have included funding for initial design and verification work for the SRAMs. Submission and irradiation of test structures would occur in future years. Deliverable: Design and simulation of a 65 nm radiation hard SRAM IP G. Active edge/3D development The most serious problem with our initial large module 3D design was the yield associated with placing ~25 chips on a large sensor wafer. A compelling solution to that problem is to combine 3D electronics, which offers the ability to provide connections to both sides of an IC, with active edge sensor processing, which allows silicon sensors to be butted on all 4 sides with minimal loss of active area. This would provide individual sensor/ROIC “tiles” which can be thoroughly tested and then bump bonded into large area assemblies. In addition, this allows connections to be made on a wafer scale, simplifying the process, allowing very fine pitch and lowering the cost. Deposit amorphous silicon Die 1 CMOS Wafer Die 2 CMOS Wafer Sensor wafer readout IC and pads 200 micron Buried oxide sensor trenches 3D bond CMOS Wafer Sensor wafer Etch Handle wafer Figure 11. Structure of the 3D/active edge SOI stack Thin, Expose top contacts, singulate CMOS Wafer Test, bump into assembly PCB Figure 12. Simplified 3D/active edge process based on plasma dicing and amorphous silicon contacts. Our current project, funded by generic R&D, is intended to demonstrate a combined 3D/active edge process using SOI active edge wafers from VTT and dummy ROIC wafers from Cornell. The process used for the current demonstration has disadvantages in the complexity of the SOI sensor wafer and the complex singluation process. Since that project was initiated we have developed a simplified scheme (figure 12) for producing these tiles where the active edge is produced by plasma dicing (a standard industry process) and the backside contact to the diced edges and the back side wafer contact is supplied by a doped amorphous silicon layer. This also allows for straightforward fabrication of thinned, radiation hard sensors on 8” silicon. We will test this process on spare test structures this year, and if successful, expect to make a proposal for application of this process in the Phase 2 pixels or silicon-based calorimetry. 4 RELATION TO EXISTING EFFORTS CMS has several alternate version of the “baseline” PS module. All of them will require the type of bump-bonded structure we will provide in Task A. This work is strongly coupled to both the module mechanical development development effort and to the electronics assembly effort at CERN. Task B represents the US contribution to the smaller scale 65 nm chip and sensor “MaPSA-lite” demonstration project. This work is considered a first step in the development of an integrated track trigger module. The MaPSA-lite will test the 65 nm front end chip and also allow some testing of the integrated system. If properly designed it should also allow us to test some schemes for inter-column communication. Task C is focused on mechanical and thermal tests needed to validate the PS module design. In particular high thermal conductivity carbon foam is an appealing material for providing the spacer between sensors and is the baseline for the rod support structure. Optimal thermal contact glues and the thermal resistance of the glue interface cannot be calculated from first principles. A thermal test setup is necessary to understand these basic parameters. This work will lead directly into system cooling tests which utilized the CO2 test system which was built for CMS pixels at Fermilab. Some advanced form of interconnect is necessary for communication between columns in the PS module (Task D). Here we can build on our extensive experience with 3D (through-silicon-via) technologies and excellent industry contacts based on our previous work on 3D circuits. We have discussed the “via-last” TSV implementation with RTI, Tezzaron, and Allvia and believe that we can develop prototypes quickly if funding is available. Our initial work on low profile wirebonding is based on discussions with Sandro Marchioro at CERN. The interposer design, although not the current baseline, has significant advantages and has some interest within CMS. The Fermilab IC group is currently working with the detector group at the Argonne Advanced Photon Source to develop a large-area interposer-based sensor array for x-ray imaging. Costs can probably be shared with this effort. A silicon or glass interposer is similarly a natural candidate to host sets of associative memory chips for off-detector track finding. Task E completes the existing effort on the VICTR chip and 3D design of the track trigger module. Task G is an outgrowth of our combined 3D work and large area module studies. Initial studies are being performed in collaboration with SLAC, Brown and UIC. If successful, the concept can be applied to large area pixelated arrays for the CMS (and ATLAS) forward tracker extension, a possible silicon-based forward calorimeter, x-ray imaging applications, and space science. 2 6 2 8 6 4 4 6 52 1/15/14 2/12/14 1/29/14 2/12/14 4/9/14 5/21/14 5/21/14 5/21/14 9/24/14 1/29/14 3/26/14 2/12/14 4/9/14 5/21/14 6/18/14 6/18/14 7/2/14 9/23/15 MaPSA-Lite MaPSA design available Design MaPSA test board Fabricate MaPSA test boards Assemble MaPSA-Lite Test MaPSA-Lite MPA component tests 3 4 6 12 50 2/15/14 3/8/14 4/5/14 5/17/14 3/1/15 3/8/14 4/5/14 5/17/14 8/9/14 2/14/16 Mechanical R&D Thermal and mechanical tests of thin epoxy layers Design, fabrication, and testing of rod support structures Continue and refine rod support structure 12 40 40 1/15/14 1/15/14 10/22/14 4/9/14 10/22/14 7/29/15 Interconnect R&D Design demonstration interposer/TSV Fabricate demonstration interposer/TSV Test demonstration interposer/TSV Demonstrate alternate interconnect tech 8 12 24 16 1/6/14 3/3/14 5/26/14 2/1/14 3/3/14 5/26/14 11/10/14 5/24/14 11/10/14 5/24/14 12 2/1/14 4/26/14 2/1/14 4/26/14 6 4 12 18 4/26/14 6/7/14 7/5/14 9/27/14 6/7/14 7/5/14 9/27/14 1/31/15 ASIC Design Design radiation hard SRAM IP Fabricate rad hard SRAM Test Rad hard SRAM 20 16 22 5/1/14 9/18/14 1/8/15 9/18/14 1/8/15 6/11/15 Active Edge/3D development Receive 3D bonded wafers Singlulate active die Probe tests Fabricate test modules Bench test modules Beam test modules Test DRIE with test structures Test backside processing with test structures 8 4 8 8 12 20 30 VICTR Chip Tests Receive Oxide bonded chips/sensors Test oxide bonded VICTR If initial tests are promising: Bump VICTR/SS to LS + Interposer* Test 5-layer stack* Beam test preparation* Beam tests* UCSB and UC Riverside participation to be determined. 4/9/14 X X X X X X X UC Riverside Receive funding Full dummy module Design dummy readout chips Fabricate dummy readout chips Design dummy sensors Fabricate dummy sensors Assemble dummy PSA module Electrical tests of dummy module mechanical tests of dummy module Thermal tests of dummy module Full module fabrication and testing UCSB End Fermilab Start UC Davis Milestone Date 1/15/14 Duration Cornell Task Brown 5 SCHEDULE AND MILESTONES X X X X X 7/2/14 X X X X X X X X X X X X X X 2/15/14 4/5/14 X 8/9/14 X 4/9/14 10/22/14 7/29/15 X X X X X X X X X X 7/5/14 X 9/18/14 1/8/15 6/11/15 X X X X X X X X X X X X 3/1/14 3/1/14 4/26/14 5/17/14 7/12/14 9/6/14 1/1/14 1/1/14 4/26/14 5/24/14 7/12/14 9/6/14 11/29/14 5/21/14 7/30/14 X X X X X X X X X X X 6 FACILITIES, EQUIPMENT, AND OTHER RESOURCES Fermilab ASIC group Six member group with extensive resources and experience in IC design. Unique experience in 3D integrated circuits. Flex Rio test stands Flexible FPGA based IC bench test system Silicon Detector facility - Large clean rooms with extensive infrastructure with CMM machines, probe stations, wirebonding machines and test equipment Test Beam - The MTest primary beamline consists of a beam of high energy protons (120 GeV) at moderate intensities ( 1-300 kHz). This beam can also be targeted to create secondary, or even tertiary particle beams of energies down to below 1 GeV, consisting of pions, muons, and/or electrons. This facility will be used to test detector assemblies Carbon fiber fabrication facility CO2 cooling test facility in SiDet Thermal conductivity test facility in SiDet Brown University We have a wide range of equipment available for the work at Brown University. We have two Rucker&Kolls semi-automatic probe stations, several oscilloscopes, including a high-end model, programmable signal sources and voltage sources, an LCR meter, Labview software and desktop computers needed to control the equipment. National Instruments Flex RIO system. UC Davis An electronics lab containing all the necessary test equipment for electronics development, including modern oscilloscopes, a spectrum analyzer, a digital analysis system and other characterization modules. The group also maintains licenses for various design software packages for pc board layout, circuit simulation, programmable logic etc. There is a 100 sq ft shielded room (Faraday cage) for testing sensitive circuits. An Indium deposition system for forming indium bump bonds. This equipment is housed in its own dedicated lab in the physics department. The Northern California Nanotechnology Center on UC Davis campus has a 10,000 square feet class 100 micro-fabrication facility that is available for use to the faculty at a nominal fee. This facility has photolithography equipment including a mask maker and a mask aligner, several stations for wet chemistry and equipment for e- beam evaporation and sputtering. A bump-bonding lab (see picture below) consists of a class 1,000 clean room equipped with several pieces of specialized equipment (acquired in 2010): 1) a Signatone probe station, 2) a West Bond ball bonder, 3) a Fine Tech aligner bonder, 4) EFD robotic dispenser and 5) a MannCorp programmable reflow oven. We are in the process of acquiring a new aligner-bonder with the help of ARRA funds. Research Devices flip-chip aligner/bonder and a probe station in a separate class 10,000 clean room. Support using highly subsidized electronics and machine shops Mechanical design Cornell University Nanofabrication Facility - The Cornell NanoScale Science & Technology Facility (CNF) is a national user facility that supports a broad range of nanoscale science and technology projects by providing state-of-the-art resources coupled with expert staff support. Simulation/Software including experience with Silvaco 3D TCAD simulation Sensor and readout assembly & testing UCSB Machine Shop Clean rooms and infrastructure FPFA and ASIC design engineering UC Riverside Machine Shop Clean room, probe station, semiconductor test equipment in Physics building UCR Center for Nanoscale Science and Engineering Nanofabrication Facility available to faculty for hourly fee 7 BUDGET AND BUDGET JUSTIFICATION FOR FY14 An additional spreadsheet is included which breaks down cost by task and funding source. 8 PRELIMINARY BUDGET AND BUDGET JUSTIFICATION FOR FY15 Describe activities foreseen to achieve the deliverables and provide a preliminary budget.