All foundries – IC and MEMS alike – provide the manufacturing capacity and process engineering to support the production needs of their customers. The foundry model is built on the economics that the scale and expense of building a dedicated fab for any one company is prohibitive, and providing a path to volume production without the full burden of those costs is a critical enabler for the formation of these startups and new product initiatives. Just as the “fabless IC company” drove the growth of the overall IC industry during the 80s and 90s, so too will the foundry model drive the growth of the MEMS industry in the 21st century.
And such growth is critical: we talk of the “sensory revolution” and the “internet of things” (see more about this topic in another blog), but the fact is that without an efficient way of getting these new companies to market quickly, the overall growth of the industry will fail to achieve its potential. Right now in the IC world, fabless companies represent approximately 30% of the total revenues, and well over half of the total number of companies. For MEMS, that number is more like 6% and while the number of MEMS startups is large, the challenge is and always has been to get these new products and ideas to market as quickly as possible, so that markets can pick the winners and losers and the pace of innovation can drive the market growth. (Source: WSTS and Yole Developpement figures.)
All processes take time and effort to build; the latest IC nodes cost upwards of $1.3B to develop with new fabs costing $7B to build (Source: Semiconductor Engineering), and the process development effort involving hundreds of engineers can take five years to complete; in the end though they have a process which can support hundreds of different designs. The unique challenge with MEMS is that we are using semiconductor techniques to build mechanical things – an orchestrated sequence of processing steps to manufacture a given product, but which may be different from the next product brought to the foundry for manufacture. The end benefit is the same: leveraging wafer level processing to reliably produce products cost effectively at high volumes. But by and large the processing steps need to be engineered to fit the needs of the product, and this is the fundamental challenge for the fabless MEMS model today.
But there are things we can do to help.
At Silex, we build our reputation on our ability to leverage our process integration experience to shorten our customer’s time to market as much as possible. Getting functional prototypes back to our customers quickly not only helps validate the product design and accelerate system integration, it is a critical enabler for market validation as well.
Your early interactions with our technical sales and Customer Integration teams are critical, consultative steps to map your product idea (or existing process) onto the process capabilities and SmartBlock™ modules in our fabs. The SmartBlock approach is the key to fast prototyping: these processing module elements are based on the hundreds of prototypes we have run in our history and represent the most well-understood and most commonly used process blocks needed to make MEMS. It provides a foundation for discussion with you on where the risks are, where the unknowns are, and what work needs to be done to demonstrate a working process flow for your product. What comes out of this is not just a path to prototyping, but a roadmap for successful process qualification and characterization, and ultimate release to production.
There are unfortunately no “process nodes” in MEMS that generations of products can leverage, and (at least today) some process tuning is inevitable for each new product design, but working with Silex gives you this fundamental advantage: being able to benefit from our experience, our capital investments, and our 100% focus on MEMS to get your MEMS to market, faster.