Roger H. Grace,
In my annual "Commercialization Report Card" market study, the MEMS industry infrastructure is one of 14 critical success factors. I am delighted to report that the grades have been very good over the past several years, earning an A- in 2010. "Infrastructure" includes software development tools, materials, process tools/equipment and, most importantly, wafer, packaging and test foundries.
For matter of discussion, we will assume (perhaps not a prudent idea) that the MEMS product has been fully defined through ample market research addressing the factors of user need analysis, competitive landscape and company core competencies. Next comes the product design as well as the processes by which the product (including the packaging and testing) will be created. The product could be designed in-house or via a number of independent design firms. Here it is important that the MEMS/package been modeled and analyzed as a system…a.k.a. called co-design.
MEMS wafer foundries come in all shapes and sizes and conveniently segmented into development and/or production foundries. Their primary purpose is to convert the product and process designs into hardware from an evaluation or small scale production level. Many universities provide this service through extensive facilities including the University of California Berkeley, Stanford, University of Michigan and Cornell. Also institutes including Fraunhofer, VTT and Sandia play heavily in this area.
Twenty years ago, MEMS foundries were not so "in fashion" as they are today. Their proliferation has been driven by the "lessons learned" from the semiconductor industry in its "fabless" and "fablite" models. Most of the MEMS startups in the past 20 years were fabless since the VC community prefers to invest in intellectual capital and not capital equipment. The "sweet spot" i.e. what they do well and want to support of many foundries for foundries varies dramatically. Selection criteria for organizations wishing to engage foundry partners also varies dramatically and is specific to the product and process required to manufacture. To help support this complex process, the MEMS Industry Group has developed their MEMS Foundry Engagement Guide that it makes available to the general public (www.memsindustrygroup.org/foundryguide).
Packaging and test are acknowledged as major cost and performance factors in the creation of a MEMS device. To meet the need of the industry, a number of MEMS packaging suppliers have been created. Another very interesting aspect of MEMS packaging is quickly emerging…that of wafer level packaging, chip stacking and TSV. Now many MEMS foundries are offering this capability with wafer "capping" being a mainstay of MEMS for many years. Once the MEMS device has been manufactured and tested, it needs to be tested in the physical environment similar to that in which it is required to operate and over its operating temperature range. Companies supplying large volume MEMS to applications including automotive and consumer electronics e. g. ADI, Bosch, Freescale and STMicroelectronics have refined this function to an art form. And what to do if you do not have in-house test capability…go to a foundry that does. Acutronic has opened up a development and test lab in Pittsburg to offer its capability to companies seeking support for their inertial MEMS parts.
A robust infrastructure currently exists and is expected to further develop to support the design, manufacture and test of MEMS devices. The MEMS industry mantra should be…"why buy the cow when you can steal the milk". The major challenge for MEMS developers is to choose the right partner and vice versa….sounds like life to me.