MEMS: A fast growing niche
01/01/2002
Debra Vogler, Senior Technical Editor
|
The market potential for microelectromechanical systems (MEMS), like everything else in high tech, has been in flux. Yet, a recent forecast (Table 1) indicates opportunities still exist across a broad segment of applications. One category, MOEMS (micro-optical electromechanical systems), is projected to increase at an annual compounded growth rate (ACGR) of 82%/year. Providing a lower-cost alternative to existing technologies is a key driver of the MEMS market. Two examples that could be considered "killer apps" are accelerometers, which, along with nozzles, were fully commercialized in 1998, and pressure sensors, fully commercialized from 1990 to the present. By 2006, seven more categories of MEMS devices are expected to be fully commercialized: gas sensors (2005); valves (2002); photonics/displays (2004); bio-/chemical sensors (2004); rf switches (2005); rate sensors (2002); and microrelays (2006) [1].
While there are similarities to traditional chip processing, there are also many variations and actual differences (Table 2). A cursory review shows diverse requirements for equipment and for an infrastructure that is not ready.
Barriers to infrastructure
Developing the manufacturing wherewithal to take advantage of the MEMS growth potential and market size will require overcoming two major barriers for IC manufacturers: the realities of their business model and the differences between MEMS manufacturing and traditional IC manufacturing.
While some technologies, like rf MEMS, can be made to fit the standard high-volume IC business model, the majority do not. A $2 million tool may not be a problem for an IC fab that has a $2 billion capital-spending pool, but that same tool may be too costly for a MEMS operation that has a $20 million budget. Amortization costs are another concern because there are far fewer wafers processed to spread out tool investment for MEMS in comparison to traditional semiconductor devices.
The opportunities for equipment suppliers appear more attractive to niche players, taking into account 1) the major role customization plays in serving the MEMS-manufacturing equipment segment; 2) the propensity for employing used equipment; 3) a market size for MEMS about two orders of magnitude smaller than the semiconductor market; and 4) the corresponding small equipment market for MEMS.
Toolset strategies
Process tweaking the usual modus operandi for traditional IC manufacturing does not work for MEMS. "Under the general heading of MEMS, an equipment supplier may be asked to etch through a 4-in.-dia. silicon wafer stack having a total thickness of 1.2mm, a 6-in. wafer having a thickness of only 80µm, or a quartz wafer 2mm thick," says Andy McQuarrie, executive VP of strategic marketing at Surface Technology Systems (STS). "To further compound the problem, etch dimensions and depths can vary from microns to millimeters."
|
James H. Smith, VP of MEMS research and development at Transparent Networks, points out that each MEMS-manufacturing technology leverages IC processing to one extent or another. For example, low-cost tools that take advantage of equipment specifications that can be relaxed in areas not as stringent as those for traditional ICs, such as uniformity and particulates, should be possible. "Most, but not all, MEMS technologies use the same photolithography equipment as the IC industry," notes Smith. "The remainder of the tools used in the wafer fabrication portion of MEMS products leverage equipment such as furnaces, film deposition, wet chemistry, and plasma etching, to varying degrees."
One segment that will require special attention is the back-end; on this point, most experts agree. Again, customization to the application is key. Unlike standard IC packages, which keep the environment out, a number of MEMS device packages are designed to let in a particular environmental stimulus, yet protect the device from dust, corrosion, moisture, and the like. Optical MEMS and gas detectors are two examples.
Most of the equipment in a MEMS-manufacturing operation is previously used, much of it modified in-house. Projects are often done using quick-turn development (e.g., masks made from high-resolution laser printers are not uncommon). There is also wide use of nonstandard materials like gold, organics, and KOH, with flexible work-flows, unlike the careful scheduling of a traditional fab line.
Tackling substrate-handling issues is another challenge with some problems that do not exist in a traditional IC environment. For example, membrane structures can be damaged by loading/unloading mechanisms, and clamping and cooling techniques often require customized solutions.
"Wafer thickness can range from 50µm to 2mm," notes McQuarrie. "Unpackaged die are susceptible to out-of-plane shock and the industry needs variable, custom wafer lift mechanisms to handle certain die types."
Still, Smith has noticed that, as the number and volume of products based on MEMS technology expands, the synergy between their technologies is increasing. "Possibly the best example of this is in the wafer-bonding and double-sided photolithography area," he says. "Plasma etching of both polycrystalline silicon and single-crystal silicon is another area where convergence is occurring."
Smoothing way to commercialization
Companies vying for a foothold or trying to maintain position in either the MEMS device or equipment markets, do not have to go it alone. Steven T. Walsh, director of technology entrepreneurship at the University of New Mexico, and president of MANCEF (Micro and Nano Technology Commercialization Education Foundation), an organization of about 500 members, consisting of approximately 90% corporations, believes that addressing the challenges of commercialization is key.
Walsh has articulated a concept he calls Triple Helix a partnership between industrial, academic, and governmental groups. It is based on the premise that disruptive technologies are often first embraced and marketed by small rather than large firms.
Maybe the combined efforts of industry, academia, and government are working. When Roger Grace, president of Roger Grace Associates, issued his commercialization report card at Semicon West 2001, he gave high marks to positive changes regarding venture capital attraction and creation of wealth that have taken place since 2000. Established infrastructure and industry roadmap categories have also shown improvement; profitability, unfortunately, has not.
Regarding profitability, it helps to have success. Grace has noted that the cost of a typical killer app device, in his opinion one that ships at least one million units/month, would have to be no more than about $5. "In general, 75% of the cost can be attributed to packaging/testing and 25% to silicon," states Grace. Since MEMS testing is extensive because it must be done at the wafer level, die level, and on individual micromachines, it seems obvious that attention should be directed at this segment to meet cost targets.
MEMS have been touted as the next wave for almost a decade. In a world where the next killer app might be disposable bio- and chemical sensors, MEMS appears poised to grow rapidly in the next few years.
Debra Vogler, senior technical editor, can be reached at ph 408/774-9283, e-mail [email protected].
Reference
- Source: S. Walsh, J. Linton, R. Grace, S. Marshall, J. Knutti (2000), "MEMS, Microsystems, Micromachines: Commercializing an Emergent Disruptive Technology" in MEMS and MOEMS Technology and Applications, editor, P. Rai-Choudry, SPIE (The International Society for Optical Engineering Development), Bellingham, WA, pp. 479-514. Reprinted with permission of Semi (MEMS/MST Technology: The Second Micromanufacturing Revolution).