Maturing MEMS Sector Moves towards Common Testing Protocols
By Paula Doe, SEMI Emerging Markets
MEMS folks haven’t typically had use for cooperating on standards efforts, with the wide variety of diverse mechanical structures they make, and the process technology IP that’s still a core competitive advantage. But the maturing sector is starting to see an increasing need for a common language of testing and characterization to help ease communication with outsiders, both outside manufacturing partners and the ever-wider range of customers not expert in MEMS.
Though no one is yet thinking about making the diversity of MEMS with a single standard manufacturing process like CMOS — which is what MEMS standards first brings to mind for many in the sector — makers are starting to argue that common testing protocols would help them talk to both their foundries and their customers, to potentially ease both the development and the design-in of MEMS devices to expand the market. Developers are looking for more consistent data on materials properties from foundry processes to feed into their design and simulation tools. Device makers are looking for common handling and carrier characteristics, and for testing protocols to test the same properties under the same conditions for spec sheets that actually tell customers what devices will really do, and for common reliability testing standards as a common baseline to prove that a device will last.
“I think we’re seeing an alignment of the planets,” notes Mike Gaitan, project leader for MicroElectroMechanical Systems (MEMS) at the U.S. National Institute of Standards Semiconductor Electronics Division. “As MEMS foundry services continue to grow, we can’t use custom test and calibration systems anymore, we need a common language.”
A Way towards Easing Design Simulation and Customer Design-in
Consistent measurement of materials properties would enable designers to better utilize simulation software to design their new devices, instead of trial and error, potentially vastly speeding the design process, argues Alissa Fitzgerald, founder and managing member of product development firm A.M. Fitzgerald & Associates. “MEMS has simulation software, but not accurate materials data,” she notes. “If we had common test techniques and standards for the data from foundries, we could get more out of simulation software.” Most useful, she suggests, from the development issues she sees across the range of customers her company has worked with, would be test standards for residual stress, Young’s modulus and hermeticity. To be useful these would need to be developed with input from both foundry engineers and materials experts, and would need to use practical production test methods, not those developed by academics with access to special equipment and little concern about speed and cost.
“What needs to happen now that MEMS is growing up is to define the terms we use to compare and specify parts and materials, on both the buying and the selling side,” similarly argues Jim Knutti, president and CEO of pressure sensor maker Acuity Inc. “I want to know just what my foundry means by, say, ‘low stress nitride.’ And I want my customers to know that my device specs mean the device will work in qualification runs as specified. Many specs are totally bogus.” Recent burgeoning demand for MEMS sensors for all sorts of new applications means dealing with many new systems-level customers who may not have any in-house expertise in MEMS. “We used to work primarily with educated customers who knew how to use MEMS. But now the business has matured to the point where new customers don’t necessarily know as much about MEMS components anymore,” he notes. “We need to make it easy to design our components into systems across consumer, medical, and process technology markets.”
All these tiny MEMS mechanical sensors can be influenced by a lot of parameters, including temperature, atmospheric pressure, package stress, and being off axis. Moreover, these parameters can interact, and everyone measures them slightly differently, often of course to show their parts to best advantage. “We need to strip naked all the assumptions and be clear what we’ve looked at, under what conditions, using some common general guidelines for specs that the customer can easily use,” argues Knutti. “MEMS has reached the stage of maturity where we’ll all gain more from the overall growth of the market than from each having our own unique testing practices.”
Baseline Qualification Guidelines May be the Easiest Target
One target area may be to simply create a standard listing of the tests major MEMS makers use today in common practice to qualify basic types of packaged devices, for the separate needs of military, automotive and consumer markets, suggests Mike Mignardi, manager of MEMS component development at Texas Instruments. He notes that despite MEMS reputed uniqueness, some 80 percent of qualification for mechanical applications can probably be standard. Just making public and explicit what is becoming the defacto common standard for qualification testing at major suppliers, what’s typically done to show the device will not fail, would build customer confidence in an established baseline process they can trust. And it would help head off requests that some MEMS makers report for unusual tests that might not make sense, like for fatigue in crystalline silicon, when that’s not an issue. “Publicizing the defacto standards doesn’t reveal any IP, so everyone would likely participate,” says Mignardi. Though startups would particularly find this baseline helpful, he points out that large companies would too when they diverisify into new MEMS applications.
A good place to start is to list all the different reliability tests companies now use to qualify MEMS devices, then to get the expertise in the room to decide by application what makes sense as the best practice, concurs Asif Chowdbury, director, MEMS assembly engineering at Analog Devices. “What’s really required for a consumer or an automotive application for example? What’s worth the cost? Today most of the OEMs have their own versions of MEMS qualification testing — there’s a different test for each different device for each different customer,” he says, and notes as well that even defacto standards keep changing with changing market shares, as, say, one company’s preferred tests for MEMS microphones get replaced by another. “We need baseline guidelines for people who aren’t so used to MEMS,” he says, noting that it’s to the advantage of everyone involved not to have to all keep re-inventing the wheel. “Industry guidelines would save both users and suppliers time and money.”
NIST will host a workshop open to all interested parties to reduce the growing wish list of MEMS test and characterization issues to the most important ones, to be held the day before the annual MEMS Industry Group members’ meeting in March in San Jose. The next step in the SEMI standards process would be to form a separate working group on MEMS test standards to further define the target area and goals, then to form a task force with a specific scope and charter.
Gaitan is currently making the rounds of MEMS industry events, holding informal workshops and talking to people one-on-one to better understand industry needs, so NIST scientists can focus their work on developing and evaluating MEMS test technologies on key issues that matter to the sector to expand the market. (Look for him at MicroTAS in October, the MEMS Executive Congress in November and IEEE MEMS in January.) “Round-robin” test evaluations and consensus on standard protocols or guidelines would then be developed through the established SEMI International Standards system.
A MIG committee is also starting work on the issue. “It’s still in brainstorming phase,” says committee chairman Mignardi. “But we’re aiming at test or quality standard guidelines.” Committee members met with the SEMI Standards MEMS/NEMS committee in July and tentatively agreed to initially focus on issues for inertial and possibly membrane sensors, as relatively mature applications with high-volume markets and many players, though microfluidics were discussed as another possible focus area. Follow-up conference calls have drawn the most participants of any MIG effort to date. Participants are starting to create a listing of the existing standards that MEMS makers actually use now. Many of these standards were not made specifically for MEMS, but could perhaps be relatively easily adapted to MEMS as a starting point.
Gaitan is also heading up a committee to write a section on MEMS for the iNEMI roadmap, which will focus on metrology — what’s done now and what will be needed to enable the devices coming in the next ten years in each of the major MEMS product sectors — to further guide the effort to establish testing standards. The committee targets a first draft by early November.
If you would like to get involved in MEMS standards activities, please contact Paul Trio at firstname.lastname@example.org.
October 5, 2010