Making MEMS Work, iMAPS Technical Seminar, Institute of System Level Integration, Heriot-Watt University, Edinburgh, 27 January 2011

Citation

Ling, J. (2011), "Making MEMS Work, iMAPS Technical Seminar, Institute of System Level Integration, Heriot-Watt University, Edinburgh, 27 January 2011", Microelectronics International, Vol. 28 No. 2. https://doi.org/10.1108/mi.2011.21828bac.003

Making MEMS Work, iMAPS Technical Seminar, Institute of System Level Integration, Heriot-Watt University, Edinburgh, 27 January 2011

Article Type: Conferences and exhibitions From: Microelectronics International, Volume 28, Issue 2

This event was snowed off, if that is the right word, back in December, whilst most of those resident in Edinburgh were snowed in. Happily, 27th January was a fine sunny day and iMAPS had a “sold out” event on their hands. Chairman Andy Longford bade everyone welcome and the seminar began with each of the many exhibitors having a few minutes to introduce their company and the services that they offer. It is always worth noting that events such as this do not take place without the support of industry and here such support was extremely strong. Nice too, to see that iSLI have a superbly well-equipped conference room with 24-in. flat-screen colour monitors on tables for each delegate so that the days of peering myopically at a distant fuzzy screen are happily facing extinction. The in-house Wi-Fi was also one which worked, unlike the convoluted and fearsomely expensive systems present in many a venue.

John Terry of iSLI spoke about Smart Microsystems, which is all about integrating CMOS with microsystems, the subject of a flagship project funded by the IeMRC. Scaling is predicted to reach a roadblock in about ten years time, and companies are asking if there is a market for IC technology from older fabs, and what can be done with “old” fabs. iSLI are answering this with work on Silicon+, which is where MEMS are integrated with CMOS; this is where MEMS are buried in wafer followed by CMOS processing, and variations on this theme including MCM (multi-chip modules) and hybridisation – bump/wire bonding with CMOS processing thereafter. A cross-section of an LC display was shown and here John described the processing for the creation of the mirrors for good pixelisation and thus definition. Another project is SCUBA 2, where the new system uses micro-machined membrane technology to thermally isolate each pixel, and the system is 1,000 times faster than its predecessor. The use of these mirrors in flat panels in cameras was illustrated, so the practical application of a product which is fully operational comes about as the result of university development. Work is also taking place on several other projects; WP1 involves power management, WP2 is on the integration of novel materials in SiC, in WP3, they are looking at non-conventional CMOS technologies including inkjet and screen printing, and in WP4, novel microfluidic sensing and actuation is the subject. WP5 covers the integration of sensors with IC technology.

Smart Microsystems is a project that targets one of the fastest growing sectors of the electronics market, involves a world leading research group with state-of-the-art technology and facilities, enjoys a strong commitment from a broad range of companies, and concentrates on integrating existing and novel technologies with CMOS to create innovative systems. It is encouraging to note that they have a number of established commercialisation routes in place. More information is available at: http://imnssm.eng.ed.ac.uk/

Dr Ian Sturland from BAe Systems showed how money could be made from MEMS. MEMS tend to come in modest volumes, in niche markets, and are used in sensors actuators and displays, and here MEMS make them smaller, lighter, faster and cheaper, but with higher reliability. The areas in which MEMS are used are numerous, but there as many design rules as there are devices, so each process flow is unique. In many areas, MEMS have wide application – in the medical, telecoms, automotive and computer fields, but in the military world the numbers are in the thousands, not the millions. At Bae, they go into niche product manufacture, as they are a systems organisation and that is what they do.

He described a micro-engineered thermal picture synthesiser which they had developed to test missile seeker systems without going to the cost of live firings. In the synthesiser, there is TPS pixel detail coming from 250,000 individually tested mirrors, which surely makes this one-off system mightily expensive, but we are assured that it was seen as very good value for money. A silicon vibrating structure gyroscope was described, development of this has taken 15 years, but there are now 17 million devices in use mainly in the automotive field on car braking systems. A silicon gyro is a classic MEMS device, and being smaller, lighter and cheaper lends itself to mass production. RF MEMS switches, fabricated filters, all are needed, and BAe works with other MEMS companies and organisations in various process steps such as powder blasting, gas processing, novel sputtered materials, deep reactive ion etching (DRIE) and wafer bonding. The MEMS world is complex, and thus it is difficult to make commercially successful, but by providing key processes to others in the MEMS supply chain BAe Systems is partnering successfully.

Stephen Duffy from Optocap spoke about the packaging of MEMS devices. His company specialises in full product life cycle from prototype to volume production. The purpose of packaging is to provide environmental protection, an optimised thermal pathway, and connectivity. Packaging is non-standard; all requirements are different, some have to have areas in contact with the environment, some require controlled pressure, temperature, humidity, etc. There may be a requirement for isolation from adhesives, solvents, fluxes, etc. used in standard packaging processes. Thus, there are two main types of package, hermetic and non-hermetic. The latter includes COB, CSP, BGA, volume assembly, MEMS dicing and die bonding; interconnection was described in detail, with interconnecting options including wire bonding, and the flip-chip attach processes. Then comes the lid attach/encapsulation process. System in package (SiP), was illustrated, and wafer-level packaging (WLP). An interesting case study was described, in which a 10×10-mm MEMS device was attached to a PCB, with various challenges, and the solutions provided by Optocap reflected their knowledge of both chip and package design and manufacture, and many years of experience.

Teradyne have doubled their turnover between 2009 and 2010, so they have to be doing something right. Chris Brown explained that what they do is test, and how have 50 per cent market share in SOC test both for memory and for system on chip, which accounts for 67 per cent of their turnover, with 33 per cent for defence and aerospace board test, commercial board test and automotive diagnostic systems. The increase is in inertial sensors (accelerometers and gyroscopes) where the market is exploding. MEMS testing comes in three test stages, the ASIC probe test, the MEMS probe test and then the final test. The key challenge is in material handling, they have special handlers move during final test. The cost pressures on MEMS devices require low-cost test for all three test stages, but high throughput and fast capacitance measurement is important at MEMS probe test. Having test efficiency at all test stages is key, and minimising test platforms within MEMS test operations gives production efficiencies.

IMEC in Belgium have a division called CMORE. Their François Iker explained how they were offering a very versatile range of MEMS, as well as design a technology platform, packaging and testing and reliability. Their technology building blocks include an Above-IC SiGe MEMS platform, metal and SOI-based MEMS, MEMS packaging/3D stacking and interconnects. With the Above-IC SiGe MEMS platform, the motivation is to develop a post-CMOS compatible MEMS platform, with a monolithic integration of MEMS and IC. The requirements and constraints include a low-temperature budget below 450°C and layers with controlled and good mechanical performances. The aim is for integrating MEMS and driving electronics thus improving performance within a smaller footprint. Imec have a wealth of experience with 3D stacking and interconnects, and the application of their platforms is varied, from micromirrors, gyroscopes, accelerometers, pressure sensors, to resonators and RF MEMS. Packaged accelerometers were explained, as were pressure sensor devices, narrow gap resonators were covered, GSA MEMS working group now formed. Imec offers all the expertise under one roof needed to build MEMS, with design and simulation tools, process capabilities (200 mm CR 24/7), characterisation and reliability, with different platforms including Above-IC SiGe MEMS, SOI-MEMS, metal MEMS, packaging, i.e. the full project from R&D to prototype and production.

Mark Begbie of iSLI told us about the use of MEMS in sensors. Back in the 1960s, inertial navigation systems were heavy and expensive, and used a lot of power, but now they are everywhere, cheap as chips, and highly integrated; they come as accelerometers, gyroscopes, pressure sensors, magnetometers, microphones, etc. Devices can be monolithic, or hybrid, but of interest for iSLI is the MEMS system design flow, in which feasibility and performance can be predicted, and simplification can take place. iSLI have developed MEMS for scanning, measuring humidity, as an accelerometer. Embedded, such processors are very low cost. Ultra-wide band is something that will radically change the world of data rates. Inertial energy harvesting, either macro or through vibration via MEMS is a developing field, a field in which a company called MicroStrain is active, who enjoy high levels of funding.

Alan Evans is from Unisem Europe, a very large organisation operating on a global basis. They offer cavity packages for volume in three types, LGA-FLP, LGA-MCP, LGA-MLP, with array processing for improved volume and lower cost, having shared production equipment with BGA/LGA. Describing the various packages, Alan explained that FLP stands for formed lid package, using stainless steel with a Cu solderable finish. There are many package options: the process flow includes wafer dice, die attach, wire bond, glob top, die coat, lid attach, laser mark and singulation. MCP is the moulded cavity package, which also has several options, including stacked die, SMT passives, and the lid can be of many different materials as required. The process flow was rather different. The MLP moulded lid package is a laminate substrate-based package with an LCP moulded lid. MEMS cavity packages that are developed in Unisem Europe can be transferred offshore to one of their Asian volume facilities. Only one set of tooling charges would be levied, and if qualified in UE they would offer FOC charge build offshore qualification lots, and FOC qualification stress testing for offshore transfers.

Taking MEMS to new markets would always be of interest, and Mark Hartree of Selex Galilieo rose to the occasion. They used to be called Ferranti. Space gyros are what they do and have done for over 30 years. The ESA called for reliable low-cost coarse sensors that control attitude, and this is SiREUS, and the gyros help keep a satellite on station. Mark explained how the gyros worked; essentially it comprises a planar ring with eight support legs in a vacuum cavity. The first gyro was launched into space in April 2010, and performance was pretty good. More testing still be done, both at ESA and at independent test house, and a flight operation is eagerly awaited.

Eric Mounier of Yole Développement spoke with authority on MEMS Manufacturing and Packaging Trends. MEMS remains a very fragmented market, with few applications having a market size above $200 million. But MEMS now deliver mission critical value adding functionality, and the time between development and commercialisation is getting shorter. In a market with a total value of $8 billion, it is the consumer sector that is the main driver. Mobile phones contain many sensors and MEMS, such as pressure sensors, proximity sensors, micro-mirrors, gyroscopes, accelerometer, micro-displays, etc. Other drivers include performance, needed in aeronautics, cost and form, and reducing thickness. MEMS builds on IC manufacturing with major differences. By 2015, the number of cumulated processed MEMS wafers will increase dramatically, and the use of DRIE components will increase in smart phones, as well. There is no roadmap for manufacturing so far, but there are different MEMS approaches, Above IC-Interleaved MEMS, and MEMS first. Wafer bonding, however, is challenging CMOS MEMS.

Lower cost, smaller devices and higher performance are the three drivers for new MEMS manufacturing and packing approaches. DRIE and wafer bonders are technologies subject to major evolution as both technologies are moving to the advanced packaging business! Recently, CMOS MEMS is likely to be restricted to very specific applications where arrays MEMS need very close electronic processing. For all other cases, it will depend on MEMS product cycle time, flexibility, cost, integration, market demand and power consumption. MEMS would benefit for some standardisation of development and manufacture, although the standardisation of technologies and products is on the way in MEMS foundries to lower time-to-market and minimise development cost.

This was an excellent iMAPS seminar in all respects, worth the wait, and delegates had the privilege of being able to see and listen to a wealth of detailed, well-presented papers from experts in their particular field.

John LingAssociate EditorMicroelectronics International