Despite the daunting challenges of getting new MEMS technologies into volume production, plenty of little companies with big ideas count on their innovations to grow new markets for MEMS. Companies with innovative technologies that could potentially significantly expand the applications for everything from oscillators and accelerometers to microfluidics and DLPs presented updates on their progress at the Microtech symposium organized by Yole Developpement at TechConnect World in California last week. Yole expects emerging MEMS to account for more than $2 billion in sales, or more than 10% of the total MEMS market, by 2015.
MEMS oscillators get market traction at last, target cell phone TCXO slots
MEMS oscillator sales are finally seeing significant growth, argued SiTime Corp. marketing VP Piyush Sevalia, noting that SiTime shipments have doubled in the last six months to now total 20 million units. He reported growing traction across a range of customers, including wins at 3 of the top 6 computer makers, 2 of the top 4 consumer electronics companies and 3 of the top 6 contract manufacturers. Though silicon MEMS devices have still taken only a tiny fraction of share from the $17 billion quartz timing market, Yole figures MEMS oscillator sales are poised for 80% CAGR over the next five years.
It took some time for the several MEMS suppliers to convince customers that silicon was a reliable alternative. “It’s very challenging to get massive infrastructure channel companies to change suppliers,” Sevalia pointed out.
Despite the presumed advantages of cost and size and ruggedness for integrated silicon devices, it turns out that the main driver for MEMS oscillator adoption may be simpler inventory management for a wider variety of features. The MEMS devices all start with the same basic resonator, which can be programmed to order to specific frequency, spread spectrum, voltage and the like for best performance for specific products, though the customer only needs to qualify and deal with one supplier. Quartz devices have to be cut and calibrated to order, then assembled, tested and aged, which can take 8 to 16 weeks. With MEMS, in contrast, Sevalia said, “We hold the MEMS and ASIC die in die bank, and assemble, test, program and ship the devices in 3-5 weeks.” He noted the value of short leadtimes for customers like one major consumer electronics supplier, whose unit demand fluctuated from 100,000 to 400,000 units a month.
Coming next are higher performance, temperature-compensated MEMS oscillators that aim to displace quartz in the TCXO market as well, particularly the giant cell phone market where MEMS has yet to match the performance requirements needed to reliably lock in on satellites and cell towers. Startup Sand 9, just coming out of stealth, said it would start sampling a chip-scale packaged device this summer with IC suppliers who can offer it with joint reference designs for their products for a solution without quartz.
Sand 9 claims to have figured out how to make a MEMS oscillator with both low temperature drift and low noise, apparently by analog tuning, and a novel architecture that involves suspending and acoustically decoupling the resonator from the silicon substrate in a predefined cavity in an SOI engineered substrate. The MEMS wafer is capped with the ASIC wafer for very short low noise circuits, then diced into 25,000 units per 8-inch wafer. First units however are put into standard 2x2mm packages to go into existing TCXO slots until the startup can prove it can meet volume requirements.
“We think MEMS will replace quartz within five years,” said Matt Crowley, founder and VP of corporate development, noting how all five quartz devices in a smart phone might be replaced with a single die. Next on the company’s roadmap is a version with the thermometer integrated directly with the resonator for perfect temperature correlation, to optimize the device for GPS applications, where the TCXO may be the most expensive and largest part of the system. Eventually perhaps all functions could be integrated into a location IC, integrating the GPS ASIC with the TCXO-grade MEMS oscillator, gryo, digital compass, altimeter and GPS filter.
Also coming next to the MEMS oscillator market will likely be some intellectual property lawsuits, noted AM Fitzgerald CEO Alissa Fitzgerald. She pointed out that the patent office is granting patents very liberally in all emerging MEMS areas these days, which results in many overlapping claims and ample potential for lawsuits, once one of the competitors starts making serious money. There are a lot of overlapping patents on oscillators, and the market may be getting big enough to start attracting litigation.
DLPs and faster light modulators for precise measurement of 3D objects
SeikoWave, meanwhile, reported interest in a wide range of potential applications for accurate, high speed 3D measurements, made by scanning a light beam across the object with a digital light projector or grating light modulator, capturing the distortions with a camera, and calculating depth by triangulation with clever software. The DLP is fast enough for most abblications, but a light modulator that electrostatically deflects ribbons to selectively let light through works in microseconds, and can be made in MEMS for a few dollars. “Far more people care about this than even I thought,” said president Matt Bellis, noting applications in facial recognition, fingerprint measurement, wafer bumping inspection, accurate measurements for making dental fillings, and measurement of breathing capacity by chest movement.
Consumer applications for microfluidics
Microfluidics micro pumps, valves and filters may turn out to have many more potential applications than just medical research and diagnostics, now that some low cost plastic alternatives are available. Bartels Mikrotechnik CEO Frank Bartels reported his company’s plastic micropump is being used not only for the typical lab-on-a-chip applications, but also for micro delivery of fluid for things like continual minimal lubrication of sensitive equipment. Interest is also coming from makers of air fresheners and continuous scent infusers for the home market –and to add scent to artificial flowers. “The Chinese flower industry is very interested,” said Bartels. “Microfluidics is not just medicine—it’s all these other things too.”
The low cost plastic microfluidic pumping system is also used for very low energy bistable electrowetting displays, which essentially pump colored liquid from one reservoir to another to change the color of an indicator, which remains that color without consuming energy until changed again. These simple indicators could replace LED indicator lights in electronic equipment to significantly reduce energy usage, as will be required by European regulations going forward.
Bartels suggests that micropumps also proves to work as effective microactuators, where pumping air or fluid into a flexible plastic reservoir to inflate it like a balloon can serve not just to open and close valves, but also as robot muscles to make things move –from phone call vibration indicators to toys to automated systems to clean the outside of windows.
Even lower-cost magnetometers and accelerometers by using CMOS
Aiming to shake up the motion sensor market with ultra low cost monolithic integrated devices is Baolab Microsytems, which aims to make its sensors directly out of metal CMOS layers, in a standard CMOS fab. “The problem is the bad mechanical properties of metal,” noted CTO Josep Montanya i Silvestre. “But it’s the cheapest way.” He said the company relies on a novel detection algorithm and autocalibration to get a good signal from the simpler device.
To keep cost down, Baolab aims to make as few changes as possible to the CMOS flow, and can get by with just three adjustments. It adjusts the silicon content and RI of the SiN, adds a HF vapor phase etch step to remove the intermetal dielectric to release the metal structure through openings in the cap layer, and then adds a sealing step to close those openings. Montanya said the design is robust enough to not be too impacted by the different properties of different foundries’ metal stacks.
The startup, which has has raised some $9 million to date, is currently working in partnership with a couple of top CMOS foundries, planning engineering samples of a 3-axis magnetometer by the end of the year, for the handset market. It targets production ramp in 2011. A 3-axis accelerometer is under development, a combo 3-axis accelerometer, 3-axis compass on the future roadmap.
Paula Doe for Yole Développement
SOURCE