MEMS fab services
Just a few short years ago the MEMS fab business was decidedly in the doldrums. Today it is catching a tailwind. Product companies are either moving toward or are already in medium and high volume manufacturing - and the latest trend is to outsource. As a result, the fabs that made it through the maelstrom are picking up new business and expanding their offerings. Four leaders shared their perspectives on making MEMS with Small Times’ David Forman.
Q: Describe the market for MEMS fabrication services. Is it growing? Static? Declining?
Vice president, marketing & business development
ALTON: We have experienced very significant growth recently - 40 percent to 50 percent revenue growth - and the pipeline of new opportunities looks very strong. I can’t speak for other foundries but my sense is that the industry overall is growing.
VCs are investing more and the failures associated with the optical bust are now a distant memory. Furthermore, the VCs are not investing in MEMS fabrication capacity. Therefore, most new start-ups that incorporate MEMS are going fabless.
HEATON: We’re seeing strong interest in our services and our business is on a fairly steep rise both from new and existing customers. The interest is from both small and large companies - primarily fabless companies or companies with fabs who require more capacity or a more complex materials set.
HERBIG: The MEMS foundry market is certainly growing, but more slowly than expected. From a technology standpoint, integration of CMOS and MEMS is hot.
Q: There are conflicting reports about what is driving new demand for MEMS fab services. We hear biotech, defense, telecom. What applications are driving your fab business?
Business development director
ALTON: It may be different for other foundries but Micralyne is experiencing very rapid growth related to products for optical telecommunications. More importantly, our customers’ customers are ordering more product so this isn’t an issue of building capacity for anticipated demand. End user demand is real.
Most of the growth we are seeing is from opportunities we would categorize as “medium volume and high value-add”.
While we are obviously pursuing the high volume opportunities, we’re very well set up to handle medium volume. That’s where a lot of growth is coming from.
GORNIAK: Semefab is a provider of MEMS foundry services and therefore is open to all companies seeking silicon processing for MEMS applications. The key application fields where Semefab is seeing growth are in microfluidics, RF MEMS (albeit still in R&D), biomedicine, accelerometers and thin film applications such as pressure sensors. The defense industry is slowly waking up to the possible applications of MEMS in security applications and is seen as a major contributor to business by 2010.
HEATON: IMT is diversified by design and we’re seeing strong interest in several areas. We’ve announced our work with Ion Optics, now part of ICx, in both the defense and industrial spaces for IR emitters and gas sensors. And we’ve announced that we are entering production for an optical telecom device for fiber-to-the-home with Xponent. We are also in high volume production for a switching application for telecom and shipping on the order of two million working MEMS switches each week for that customer. Other areas showing strong potential are IR imaging, RF and MM-wave, and biomedical applications.
HERBIG: We see significant activity in microphone, inertial sensor and bio applications in the telecom, automotive and medical markets.
Q: What are you doing to facilitate process development and transfer, especially for clients who have developed prototypes on equipment that differs from yours?
Vice president, marketing & sales
Innovative Micro Technology
Santa Barbara, Calif.
ALTON: This is a very common issue as many of our customers will - before they come to us - create a process or design or fabricate a few prototypes in an open access or academic fab.
Our more mature customers however will contact us first before designing their products in detail. They realize that if they design with our processes, their development and time-to-market costs can be reduced and production yields will be higher. This would seem common sense but few companies actually do it.
GORNIAK: MEMS products rely on custom processing. Therefore every new MEMS inquiry demands a new process flow. In many cases the products require an investment in new equipment. This has obviously to be balanced with the potential value of the project. It is a fine line that has to be trodden between accepting new business and the return on investment of new equipment.
HEATON: Our customers range from those with a spec to those with a design, to those that have prototyped or even produced working devices. In all cases, we collaborate closely on both process transfer/development and design for manufacturability. Our skills and experience in these areas enable smooth startup or transfer to our facility of processes even where the tool sets do not match up. Tight collaboration is key and several of our customers have personnel permanently or periodically assigned at our facility. We believe in strong and open partnering as the solution.
HERBIG: You have raised an intriguing issue because the first law of MEMS is “one product/one process.” Companies need to offer a wide variety of process options. X-FAB is continually expanding its portfolio, and is investigating or implementing several new modules, including DRI etching for bulk micromachining, electroplating, and wafer-level packaging/encapsulation.
Q: What technologies are you using to lower cost and increase reliability?
Product marketing manager
X-FAB Semiconductor Foundries AG
ALTON: A couple of things we are doing are to add higher capacity tools along with focusing on lean manufacturing. These are having a significant impact on cost reduction once a process has been stabilized. Unfortunately, for some key processes, such as lengthy DRIE etches, few if any multi-wafer, high volume production tools currently exist.
Packaging and test are two areas that have a huge impact on cost and may be part of the services offered by a MEMS foundry. As compared to 4 to 5 years ago, more attention is being placed on these areas.
Finally, the possibility of setting up a fabrication facility in a lower cost environment is coming up more. We are looking at this and the opportunity to reduce labor costs is massive. Having said that, the issue of IP protection has to be carefully addressed. I think it is inevitable that we will see more MEMS production transferred to Asia.
GORNIAK: We cannot answer this just now.
HEATON: Our team and facility were built for production, including SPC/SQC and what we call product engineering. This latter function comprises a team of engineers whose specific role is to identify and create paretos for the key parameters that drive product yield and reliability. These parameters are control-charted and tightly monitored with metrology and a variety of functional and electrical tests. This information is continuously applied to increase yields, decrease labor content and increase reliability - all of which drive down costs.
HERBIG: X-FAB aims to develop batch processing for MEMS, and looks for synergies between CMOS operation and MEMS operation, such as doing MEMS wafer batch processing instead of single wafer processing. We use a KOH batch etch instead of a DRI single wafer etch whenever possible. X-FAB also reduces packaging costs by providing wafer-level packaging.
Q: How do you compete with the big foundries that have guaranteed lines of business from parent companies that support their infrastructure?
ALTON: We don’t see competing with these companies as a huge issue and, in fact, we see a trend for them to outsource more to companies like ours.
In terms of their foundry services, the customers we talk to say they have challenges in dealing with these companies. For example, these mainly captive fabs are not able to deviate much from their internal processes or might not have the flexibility to address issues that arise during the early, less defined stages. There’s also a concern that external customers will be treated as second class citizens compared to internal product lines. While we can’t ignore these companies, we believe we can compete with them very effectively.
GORNIAK: Semefab belongs to the Semelab group. However Semefab acts completely independently. Semelab is not involved in MEMS and thus we do not get the benefits of having a guaranteed line. Semefab is quite confident of being able to compete with the big foundries as there are many niche market products appearing for volume production. These all require custom processing and hence the major manufacturers will not get involved in the small initial volumes.
HEATON: Our business growth demonstrates that we are competing very effectively and offering exceptional value to our customers. We’re not so small ourselves with over 30,000 square feet of active fab workspace - and we believe we also offer a unique mix of capabilities not offered elsewhere, including strong collaboration from design to prototyping through volume production. Our product and customer diversity also gives us a broader and growing range of experience that can be applied to solve our customers’ problems. We also continue to see the large foundries with their own products having diminishing interest in their external customers as their internal business ramps up.
HERBIG: This is not an issue for X-FAB because we have exactly this type of support.