STUTTGART, GERMANY---How long has MEMS/microsystems been (one of) the next big things in microelectronics? At least as far back as the late 1980s, which is when I began covering the semiconductor and related microelectronics manufacturing businesses.
Until recently, MEMS' projected takeoff has remained, just that, projected, with the market seeing steady but unspectacular growth. But several factors have changed to pull in the likelihood of that long-forecast boom, primarily the dawning of a consumer market for microelectromechanical systems (combined with ever-more automotive-sector apps) and the concurrent push to volume manufacturing.
Several presenters at today's International MEMS/MST Industry Forum, held in conjunction with Semicon Europa, offered compelling reasons for being bullish on the sector. Jiri Marek, senior vice president for MEMS leader Bosch, provided an informative overview of the company's activities and the sector in general. He said the company maintains a wide lead in the sensors sector, earning more than $300 million, about double the next ranking competitors, BEI and Freescale. Bosch expects to manufacture nearly 160 million MEMS units in 2007, according to Marek, the continuation of a strong, multiyear upward trend, a trend likely to continue and even accelerate.
He detailed the company's 200-mm wafer-size-conversion project. The fab, scheduled to begin production in 2009, includes 4600 square meters of cleanroom space and will hit 800 wafer/day capacity output. Advanced CMOS, HV-CMOS, and MEMS processes will run in the new facility, initially down to 0.8 micron with plans to push to 0.18 micron, according to Marek. The German firm estimates it will spend around 600 million Euro for the factory, not the stuff of $3 billion megafabs but a sizeable chunk of change by MEMS standards.
The spinout of the Sensortec division represents Bosch's efforts to proliferate to the consumer side its considerable acceleration and pressure sensors expertise, which it has garnered in the automotive application space. With consumer apps come volume demand: "You have to have production volumes with devices to justify the efforts," Marek insisted.
Despite Bosch's sincere move to 200 mm, it lags behind STMicroelectronics, which is already well along in its ramp of its new 200-mm fab in Milan. STM's Benedetto Vigna spoke on what he called "the MEMS consumerization wave." His company has become synonymous with a MEMS consumer product, namely its 3-axis accelerometer which gives the Nintendo Wii much of its wide (and wild) appeal.
Vigna identified several emerging MEMS market apps, each with the potential for 100 million to more than a billion pieces unit-volume per year. The list reads like a gadgetheads tote bag: Mobile phones, portable media players, laptops, game consoles, hard disc drives, and digital cameras/camcorders. "Motion sensors are revolutionizing the way people use handheld devices," he opined.
In his comments about the new 200-mm fab, he pointed out that STM's was not an older, repurposed facility but one purpose-built for MEMS. Besides the 2x or better cost reduction gained from going to the larger wafer size, Vigna noted the reward of best-in-class equipment supplier technology and support, something ever-more-difficult to receive with 150-mm equipment sets, since few companies do their advanced work on the older toolsets. Among the manufacturing challenges cited were deep silicon etching, wafer-to-wafer bonding, sacrificial oxide etching, and wafer thinning in the front-end processes, and packaging miniaturization and testing-time reduction with the same accuracy in the back end.
During the Q&A following Vigna's talk, someone asked about the possibility of another wafer-size transition---to 300 mm. (Geez, the company hasn't even caught its collective breath from the 200-mm move---Vigna still seemed like he hasn't caught up on sleep from the ramp---and someone's asking about 300 mm!) The STM exec pointed out issues of tricky critical dimension control with bigger wafers as well as the prohibitive costs of fab conversion, but then zeroed in on a very MEMS-specific determining factor. He said that it doesn't make sense to move to the larger wafer size, in terms of economies of scale, until and unless the minimum feature sizes dip below 100 nm. Since STM's current MEMS device features range from 1 micron to a shade over 100 nm, the answer will be "no" for the foreseeable.
I do have one bone to pick with Vigna, Marek, and most of the other presenters: Why do they continue to use an English measurement unit---inches---to refer to wafer size? Here we are, smack dab in the middle of the EU, and European professionals continue to trot out a very American metric. They don't use square footage in notating fab cleanroom dimensions or ounces in denoting liquid volumes, let alone Anglicizing the micron/nanometer designations. Why not make the complete wafer-size transition once and for all and convert all your reference materials to the metric system?
Old habits die hard (in this case, very hard), but when it comes to wafer sizes, the millimeter should rule.
|
Blogs 
