by James Montgomery, News Editor
A new report from SEMI and Yole Developpement Group finds several emerging trends in microelectromechanical systems (MEMS) that promise new opportunities for chip companies and suppliers to expand their business. Integration of CMOS continues, with synergies in technology and processes flowing both ways. New supporting technologies are taking shape, particularly for packaging materials, and large chipmakers (and many foundries) are becoming increasingly active in MEMS development, and giving the MEMS a shot of legitimacy.
The worldwide market for actual MEMS devices is around $5.3 billion, with a projected 13% CAGR to about $9.86 billion by 2010, while sales of MEMS-enabled systems (e.g., airbag systems, and flat-screen TVs with Texas Instruments’ micromirror “DLP” technology) should double to about $95 billion by 2010. These numbers are dwarfed by semiconductor industry metrics, but offer much higher growth prospects — e.g. the market for MEMS silicon microphones is seen topping nearly $400 million by 2010, vs. $65 million last year. (SEMI and Yole define MEMS, aka “micromachines” in Japan, and “microsystems” in Europe, as microscale devices with a physically moving part, or fluidic system that sends/receives a signal. Not included in the SEMI/Yole report are things like thin-film heads for hard disks.)
Supporting the MEMS industry is a ~$1 billion equipment/materials supplier base, roughly two-thirds ($631 million) for equipment. The MEMS equipment segment is seen with only 6% CAGR through 2010 (to $861 million), but MEMS substrates and chemicals should enjoy 15% CAGR to $771 by 2010. Frontend manufacturing equipment accounts for 53% of the MEMS equipment market today, with assembly/packaging equipment taking up roughly a third, and the remainder in R&D tools. Processes such as lithography, deposition and etch are well-known to MEMS makers, with a large number of products using to create very deep trenches with high aspect ratios. Familiar semiconductor suppliers with MEMS businesses include Suss, Lam Research, and JSR Micro (for photoresists).
Etch, wafer bonding, assembly/packaging, and test equipment are expected to see the highest growth over the next five years. Most of the tools used in MEMS today are 150mm, but increasing adoption of 200mm equipment is expected for certain types of processes, e.g., wafer bonding and deep reactive ion etch (DRIE). A lot of test equipment is also modified in-house. The MEMS materials segment should see 17% CAGR to $570 million in 2010, with see eye-popping growth in coatings used for packaging and as sacrificial layers, projected to soar from just $400K today to $3 million by 2010 (e.g., IBM’s SU-8 technology, being marketed by Micorchem in the US and Europe’s Gersteltech). Increasing use of DRIE means more etch gases will be used, though the small MEMS chemicals/gases segment will double to just about ~$66 million by 2010.
Global markets and forecasts for MEMS devices, materials, and equipment
Markets……………………2005 (US$)………….2010 (US$)……….Compound annual growth rate (CAGR)
MEMS materials…………$385 million……….$771 million……………15%
MEMS equipment……….$631 million……….$861 million……………6%
MEMS devices…………..$5.3 billion………….$9.9 billion………………13%
MEMS systems…………..$48 billion………….$95 billion………………15%
Source: SEMI, Yole Developpement
Most MEMS devices are manufactured on silicon, using the same processing technologies and equipment/materials as semiconductors. Early MEMS devices were used primarily for defense and university work targeting small niche applications, but now the segment is seeing significant growth as semiconductor manufacturers see the synergies, and potential for higher yields and volumes, according to Lubab Sheet, senior director of emerging technologies, SEMI. “The technology is flowing from semiconductors to MEMS, and MEMS is finally embracing it, instead of begrudgingly.”
Another growing trend is integrating CMOS with MEMS, leveraging semiconductor manufacturing technologies to drive down costs and improve output yields. While many of today’s CMOS-based microsystems still require the insertion of dedicated micromachining modules in between the regular CMOS process steps, there is a clear trend toward not interrupting the CMOS sequence at all. Another bonus for taking on MEMS work is that they can leverage fully appreciated production lines, a far more financially attractive plan than say, TI investing in a new 300mm line just for DLP technology.
This is drawing interest from big chipmakers, including the likes of Intel and TSMC, although the amount of their business from MEMS is still very small. “When you can say TSMC, Lam Research, and Intel have MEMS-related programs and are developing MEMS products, it makes people pay attention,” said Sheet, pointing out that TSMC was a headline participant in SEMICON Taiwan’s MEMS forum. Curiously, though, she noted that both TSMC and Intel, which had been touting MEMS as a complimentary technology for its life sciences and chip integration efforts, have recently become more tight-lipped about their MEMS involvement.
Foundries in general are starting to make a name for themselves as MEMS producers, Sheet noted. SiTime’s silicon resonator, seen as a lower-cost replacement for quartz oscillators (potentially a $3 billion market, according to the fabless firm), and Acoustica’s CMOS-compatible MEMS microphones, are made by foundry partners.
The next breakout segment for future growth in MEMS is microfuel cells, with a projected market entry point in 2008, according to SEMI and Yole. Electronic devicemakers such as Nokia are keen to drive costs down and keep yields high, while addressing the usual market challenges faced by any new technology — make it smaller, consume less power, and reduce heat emissions. Sheet related a conversation with an exec at Motorola, which is trying to come up with a low-cost fuel cell for its cell phones. “He was not a semiconductor guy, but saw clearly that there are semiconductor manufacturing techniques that are proven to drive out costs and achieve higher yields,” she said. “It’s really changing their approach to developing microfuel cells to use semiconductor technology.”
On the semiconductor equipment side, technical challenges receiving particular attention these days are in MEMS assembly, testing, and packaging — moving parts of 3-D structures have to be hermetically sealed and at low cost. Due to the wide variety of different MEMS device configurations, and a lack of available solutions on the market (i.e., a business opportunity for suppliers), much of the work in MEMS testing and packaging is happening in-house at the MEMS manufacturers themselves, and is leveraged as a source of competitive advantage — e.g., Texas Instruments has tweaked some testing equipment in-house to work with its MEMS devices such as DLP. Solving these challenges for 3D MEMS devices will likely pave the way for flowing technology back into the semiconductor industry, which is still early in its with 3D structures and interconnects. — J.M.