Isn't the wafer capped, to keep out contamination?
ADI: Yes. Everything we do now is capped. But, historically--for over 10 years--we put the die into a hermetically sealed ceramic package. But now we put a cap on it. We have a second wafer that has a structure with a cavity [for the die] and a ring that goes down on top of the circuit. We use glass that flows and melts around 400 degrees Centigrade. It comes down aligned on top of the regular circuit wafer with the MEMS, where we heat it up so it bonds and forms a hermetic seal.
EET: It sounds like you've taken a normal macroscopic hermetic-sealing method and "MEMS-ized" it?
ADI: That's quite correct. We wanted to take it down to the individual die level--like a little clamshell right over the MEMS portion, so you can then treat the die just like any normal die. Because you have protected the MEMS part, you can then wire-bond--put it in a lead frame and shoot it up with plastic.
EET: That has really got to lower the cost too.
ADI: That is what we are striving for. It cuts the cost and makes a smaller part, which makes it easier to handle for all the new cell phones and video-game controllers.
EET: Yes, cell phones and video-game controllers are your newest application areas. What would you say were the major milestones in your MEMS efforts?
ADI: Just getting something to work, in the late 1980s and early 1990s--it was not as easy as we first thought, because of all these "stiction" [static friction] issues. We never before had to deal with very fragile mechanical structures on the top of an integrated circuit. We had to build special tools to manage it, and it took three or four years of process development just to get all that to work together--that was our first major milestone.
Then we had to get some customers, so we started with the toughest ones: automotive. We needed a clear, large market--which was the airbag market. We needed an application that was large in volume and in dollars, with good growth
Potential, so we could support the investment we had made.
Getting our initial customers around the 1993-94 timeframe was the next big milestone. Basically, we got GM [General Motors] and a few others to take a chance with us. Then because we promised such an advantageous price/performance point, a lot of them then switched over to us. In fact, after everybody started signing on we started running out of capacity. So we brought online a facility in Cambridge [Massachusetts] that was dedicated just to MEMS. That was our third major milestone--we brought a couple of hundred engineers together in Cambridge, where all they did was MEMS. And, eventually, we had to also build a dedicated assembly and testing facility in Amkor [Singapore.]
The other interesting thing here was that the founder of Analog Devices, and the chairman of our board, Ray Stata, was the general manager of the Cambridge facility at that time. He had a vision from the early days that addressing automotive applications would really give Analog Devices a push into the high volume, high quality, low-cost marketplace. We had a unique technology, a great opportunity, and we won a lot of business with our dedicated facility--the only way we could have met the automotive requirements, which are much more stringent than those you're typically faced with for industrial applications.
EET: Yes, that's for sure. Automotive applications are almost the definition of visionary.
ADI: Yes, I think so. And it was tough going for many years, because in automotive applications you have to first prove you can do what you say you're going to do. Then, the next year, they give you a little more business. And if you do well again, then the next year you get a few more customers. It was slow going, but Ray [Stata] stuck with us by basically managing MEMS as a separate entity within Analog Devices--that gave us enough breathing room to build the business. We were not profitable for the first five or six years, because we had to grow along with having a low-margin, high-volume business.
EET: What is your most important recent milestone.
ADI: That would be the three-axis accelerometer--I'm sure that your EE readers are very interested in that chip. We have had a two-axis device for more than ten years, but we had to hunt for very innovative ways to get the third axis out of an existing two-axis structure.
EET: That would be a three-dimensional structure?
ADI: Yes, it moves up and down as well as side to side, but the trick and challenge was to pull off the electrical signal as it moved up and down.
EET: Because the throw is not as far up and down?
ADI: Yes, it's not as far, but, more important, the signals are correspondingly smaller, too. Most of our competition had to add a second sensor to get the vertical, or z-axis--what we call a trampoline or teeter-totter structure. Instead, we were able to do it entirely within the same mechanical structure, because we are the only ones who can fabricate the electrical and mechanical structures on the same CMOS chip. Because of that capability, we can pull off very, very small capacitive signals from the z-axis--down in the atto- and zepto-farad range [atto is one quintillionth, or one billionth of one billionth (10 to the minus 18th), and zepto is a thousand times smaller yet, or 10 to the minus 21st].
EET: Holy cow! So are the three-axis accelerometers going to be used only in new and different applications?
ADI: Yes, that is really what has begun to happen in just the last year. Engineers are starting to design-in our three-axis accelerometers for all sorts of new application areas. Nintendo started this by putting our accelerometers into their handheld games, and just last month they came out with their Wii system, and we are the accelerometer in the main controller. By now, they have shipped millions of units using our three-axis accelerometer.