Pity the poor MEMS developers, pushing what the industry sees as an emerging technology. "We can give you miniature micromachined components, with the benefits of lower power, lower cost, and reduced real estate," they say. "All you have to do is
change your processes and architectures." ALERT! ALERT! Seriously, is this any way to convince mainstream electronic equipment manufacturers to adopt a new technology, no matter how beneficial it may be?
We can't ask the industry to proceed full-throttle to revamped architectures and complex processes that in some cases are anathema to traditional IC processes. Despite the benefits promised by microelectromechanical systems (MEMS), we need to offer the industry a transitional solution to get it to feel comfortable with the technology and see the advantages.
So what's a MEMS system house to do? Here's a new concept. Let's figure out a plan that fits into the existing technology infrastructure and aim for a large, established market. Why? Doesn't all the entrenched competition make that the toughest route? In fact, no. Huge anticipated volumes are necessary because we must motivate our potential customers and manufacturers and let them know that any efforts they make in this area will be worthwhile.
As a fabless semiconductor company, you need to work with these big customers and their technology infrastructures in such a way that you help them to successfully ramp up and address technology issues. In turn, these semiconductor providers will be your best advocates, offering your technology to their stable of major customers as an option for volume manufacturing.
But this still doesn't mean massive implementation of your MEMS technology. Pushing technology that enables an IC with five on-chip RF-MEMS components is too scary—too many variables, with the potential for massive system failure. It's hard enough for mainstream fabs to make the transition to copper interconnect and low-k dielectrics, let alone adding complex MEMS to the mix.
So let's go with something we know. We'll get the customers to evaluate a single miniature component. We won't even call it MEMS! We'll call it a "microcomponent." Maybe it's a direct replacement for your standard, bulky old quartz crystal oscillator. But it's only 1 by 1 mm and handles multiple frequencies. In volume it can slash as much as 30% of your crystal oscillator cost. "Hmmm," the prospective partner will probably say, "I guess I could test-drive that."
"Not only that," you add, "but our components are simple and they're manufactured using standard materials and processes to ensure high yields. So you can manufacture these things yourself while keeping costs low."
Once the component is part of a reference design, you dramatically increase your partners' comfort level while keeping it within the bounds of their existing infrastructure. In addition, the partners are happy because they get to help define what may become mainstream technology, and they think they're on the cutting edge. You don't have to tell them this is the baby stuff.
All right, so it's not quite this simple. Some other considerations need to be addressed. MEMS, as "microelectromechanical systems" implies, is a system. This means stuff goes with it, like advanced electronics and packaging.
But if your head's screwed on straight, you'll know that these are just additional ways to help your customer/partner. The electronics can gain you major versatility in areas like frequency variation and enable you and your customer to control the technology by design rather than by the material.
Packaging is equally important. Prior to approaching your customers, you can manage costs (they'll love you for it!) by using standard packaging for your microcomponent. Maybe it's a little bigger than you'd like for a miniature device. But you know what? It'll get you in the door.
Reprints
