If you want to bring the conversation in a room of MEMS suppliers and end users to a fevered pitch, just say the words, "MEMS standardization." Based on the exchanges among MEMS suppliers and MEMS end users during MEMS Industry Group's Pre-conference Symposium on MEMS commercialization and product realization at Sensors Expo 2012, the room temperature will increase about 20 degrees and anyone who was checking email on their iPad will suddenly begin to pay close attention.
So while it's not news that standardization allows designers, OEMs, fabs, and foundries to leverage the progress of others and to avoid reinventing the wheel, it is still a complex, multi-layered and controversial topic within the MEMS ecosystem.
In an ideal world, standards can address processes, design tools, simulation tools, packaging, prototype qualification, production test processes and procedures, data sheet formats and conditions, as well as non-electrical performance assessments (pressure acceleration, orientation), and environmental maximums. You see such standards with semiconductors, but with MEMS, no such standards yet exist.
Even the simple word "standardization" has many dimensions. You can have a structure where internal functions and operations (such as modeling tools and software algorithms) are proprietary and customized, while the handoffs to the next step (for example, your design files) are based on accepted industry standards. This gives you the benefits of standardization without being boxed in or having your creativity limited.
But despite its presumed virtues, reliance on standards has its consequences: it can dull the proprietary edge that most vendors claim to want. It also forces designers and marketers to be specific when they announce what they really bring to the party, who they want to be, and why they should be in the top tier of those being considered for that next design-in or innovative application.
Standards can also become a limiting constraint on progress and innovation, instead of a liberating one. In the semiconductor arena, an industry roadmap has guided progress in process size, wafer size, fab technologies and test—for decades. The very public promotion of this roadmap has also tended to suppress "out of the box" ideas that are not compatible with the planned path.
We also know that some of the greatest and most profitable innovations have come from those who retain flexibility in their approach to industry standards. If there's a better way than the 'standard' way, they choose what's better. MEMS can learn from this approach.
"They said it wouldn't and couldn't work, so we did it anyway" has been the story of many successes that were considered at first to be outliers, or even radical. Standards that are set too early in an industry's life cycle tend to be straightforward, obvious extensions of what we can see and where we hope to go; too bad they are rarely where we end up. Any discussion of standardization must focus on the market rather than the technology itself.
Rather than force the choice between 100% standards or 100% proprietary, we're most likely heading toward a hybrid approach that balances standards and differentiation, with each vendor choosing the mix that makes sense for their application and expertise.
It's probable that the MEMS industry will follow the path of so many other areas of IC and end-user product advances, where standards or standard modules will cover many aspects of the design and fabrication sequence. The majority of vendors will adhere to them most of the time to get maximum leverage from broad-based resources while reducing the risks and the workload they have to undertake on their own. At the same time, they'll deviate when they think it makes sense and can give them a development or market edge. Very few vendors want to be stuck in the commodity niche, if they can avoid it.
Other vendors will take a big chance, hoping for bigger rewards. They will choose to shoot for maximum performance or functionality—focused on a few critical parameters, such as accuracy, power efficiency, or dynamic range. To achieve differentiation, they'll be more willing to go out on their own for more of the product and process development, as well as the production and test.
For MEMS end users, and even for some MEMS device manufacturers, even small steps toward standardization are welcome, as it will help to further grow the MEMS industry. During the MEMS preconference symposium, Steve Whalley, sensors director at Intel, stated "MEMS are becoming more prevalent in CE (consumer electronics) products, but just integrating MEMS in phones, tablets and PCs is complex. There is no 'MEMS for Dummies' book as far as I know, and if we are to scale MEMS across existing and new CE products, we need baseline standards in certain areas—even just standard data sheet parameter specifications would be a great start."
MEMS Industry Group is a strong supporter of providing the right environment for standardization without forcing it, as we must still allow for differentiation. In 2011, for example, we partnered with National Institute of Standards and Technology (NIST) on a MEMS Testing Standards workshop. While we published a complete report for our members, the executive summary is available to the public.
We are continuing the conversation on the standardization of testing, but we expect that progress in this area will be accomplished via small steps. In the short term, we would like to see some standardization of data sheets. This would allow end users to conduct apples-for-apples (or accelerometer-for-accelerometer) comparisons, satisfying the needs of companies such as Intel, Dell and many others that are embedding MEMS by the hundreds of millions into consumer electronics, automotive, and other applications.