SEMICONDUCTOR VENDORS are in either in the microelectromechanical systems (MEMS) sector or will be out of business; that was the stark assertion that came out of a panel session at the recent Globalpress Electronics Summit in San Francisco.
“The interesting thing about MEMS is that it is part of an evolution that started 200 years ago with the Industrial Revolution where energy was converted in to motion. The next major step was computation, the invention of the transistor. The third major revolution with the machine is all about sensors,” said Vijay Ullal, group president at Maxim Integrated Products Inc. (Sunnyvale, Calif)
“It you can combine all three – motion, computation and sensing – you could have a truly intelligent being that, maybe, could rival of human beings,” added Ullal (pictured right). “MEMS is not just a new gadget but the third massive transformation of our society. Any company that is in semiconductors now must do MEMS or go away.”
“Maxim likes the MEMS market because it is really fragmented,” said Ullal. “Lots of customers and you can differentiate your products, which fits in with our model. We are beginning to enter the MEMS market and are starting out in areas not dominated by the established players. We are in to resonators and micro relays and eventually we will expand beyond this. It is not the MEMS device itself that is of most interest it is the combination of the device, the signal processing and algorithms. The reason we are getting interested is, it is a market with tremendous growth.”
Asked by EE Times if Maxim would look to acquire to expand its MEMS activities Ullal said that it would probably be better to find a good partner rather than acquire, as acquisitions can be quite disruptive. “The last thing the MEMS industry needs right now is to lose its entrepreneurial spirit. You see a lot of progress being made by little companies and they can be good at innovative things.”
A more long-standing member of the MEMS industry is Analog Devices Inc. (Waltham, Mass.) which has been working on products for around 20 years. Mark Martin, vice president and general manager of the MEMS division at ADI explained how there has been an explosion in applications “not just automotive and consumer but now moving in to industrial and medical and instrumentation.”
Eric Eisenhut, vice president of sales and marketing at Kionix (Ithaca, NY), which focuses on inertial sensors, explained that, “The user interface experience, if it is done well, is fundamental to the product. You have to understand the fundamental physics and mathematics that underlies sensing and we don’t expect our customers to understand that.”
A major concern for potential customers has been a perceived high price that MEMS demand. “As the devices go in to higher volume consumer based devices there is an need to bring the cost down,” agreed Scott Smyser, vice president and general manager of VTI Technologies (Vantaa, Finland). “We have got rid of the external packaging and put the MEMS directly with the conditioning ASIC flip-chipped together.”
“The manufacturing process is clearly relevant but it is also important to understand the application the device is going into,” added Eisenhut. “We don’t want to commoditize, relying solely on the delivery and price of the component – but by understanding the application integrate associated functionality on to the ASIC,” he said.
“Historically we have used quite complex processes that have run in lower volumes,” said ADI’s Martin (picture right). “That has clearly changed and I think we have as an industry been quite successful at reducing costs when you look at the volumes that are run today. The processes are maturing and the yields, that have been a significant challenge in MEMS, are coming in line with those associated with traditional semiconductors processes.
There is a learning curve associated with using the technology. Some customers have previously just bought three axis accelerometers on price. Customers are finding out that not all devices are created equal; the user experience can vary dramatically depending on features and precision of the sensing element. Over time as the market matures there will segmentation, a very low cost market as well as other segments as the product portfolios proliferate.”
So is size reduction an opportunity? “Moore’s Law will not work for MEMS,” said Ullal. “You have a certain mass and you cannot shrink it beyond that. It should be pretty clear to everybody that Moore’s Law, if it is not dead all ready is going to die. The way to add functionality to a system is through 3D integration.”
There is a perception that the MEMS industry is being led by its customers, rather than taking the normal semiconductor industry route of pro-actively promoting pervasion by showing what its products can do. “We have been a bit slow,” said Martin. “If we want the market to grow we have to help our customers, providing algorithms and software. Education is a big area for us. We explain the five motion senses in a way to simplify and then educate in each area and highlight how people can make use of them. It is not going to be a one-way, vendor-educate-the-customer process. We have realized that we have to play our part by providing tools, but we will continue to learn from our customers.”
Smyser added: “We are all looking at algorithms, software and system development around our sensors. We have done that with sports and fitness and are doing it navigation and GPS. However there are so many applications out there that it is really our customers who are driving us. It is hard to keep up with our customers at times and this is proliferating out from the consumer space. There are some many apps for the iPhone that use the built-in accelerometer that customer are looking to use similar functionally on their equipment.”
Ullal explained that if you look at the history of the semiconductor industry it was not till 20 years after the invention of the transistor that the real applications started to exploit it. After 20 years of development the MEMS sector might not be in a bad position. “A difference between the semiconductor industry and MEMS is that there has not been a platform MEMS technology on which you can base a number of different devices. In order for MEMS devices to achieve a lot of penetration you have to provide a system-level solution. The technology that is critical to make that happen is packaging. You have to provide a solution in a single package,” said Ullal
Microvision (Redmond, Wash.) is a customer for MEMS. The company has developed a projector for use with mobile devices that combines three lasers with a MEMS device and which is due for launch mid 2009. Ian Brown, vice president of sales and marketing at Microvision, said: “We have come up with the novel application and work with the suppliers to come up with a solution. It is more challenging for us to manage the potential ramp up volumes we might need. When a handset company launches a new model it expects to ship 2 to 5 million units. If we get a design in we will have to ramp up in a short time.”
Martin explained that a popular topic for the MEMS Industry Group (MIG) is standardization. “The industry is still on a learning curve. The foundries are working to come up this curve and I think we are still several years away from a standard process for MEMS. We are many, many years away from the level of broad standardization that the semiconductor industry is accustomed to in CMOS technology.”
Eisenhut (picture right) said that he sees more standardization in the applications but not the devices and added that one of the reasons that the industry has grown as rapidly as it has in the last couple of years is that there is some commonality in use of the standard semiconductor processing tools.
“The equipment set is common and mostly we are looking at standard packaging. Kionics has chosen to have in-house manufacturing and this is why we have been able to compete with much bigger companies for the tier one consumer electronics market. Over time this might change.”Martin agreed in that standardization will involve interfacing and protocols. “As an industry we need to take advantage of these but we are better not to constrain the industry but provide differentiation that will enable applications.”
So can foundries be as successful in the MEMS sector as in general semiconductors. Maxim’s Ullal said: “You really need volumes if you want to do a foundry model. The volumes aren’t there for MEMS devices and there are too many variations for the foundry model. I am sure in the next ten years there will be at least one or two versions of inertial sensors that will be done in a foundry but right now if you want the fastest turns, you have to do it yourself. You end up teaching the foundry how to do what they claim they already know.”
What comes after MEMS? “The interesting application for nanotechnology will be really minute sensors, said Ullal. “MEMS might be the really hot thing for the next 5 to 7 years but in order for us to really get scaling, it’s going to have to be with nanotechnology and that could be ten years out. What you see in the labs with self-assembling structures in 10 to 15 years you will see in really miniature sensors running on virtually no power at all.”
“If you are not in mechanical structures now, you will miss that nanostructures of the future as you have to understand who the customers are, what the applications are and how to create this industry,” concluded Ullal.
This story appeared in the April 2009 print edition of EE Times Europe European residents who wish to receive regular copies of EE Times Europe, subscribe here.
See other stories from this issue here.
You can download a digital edition of the latest EE Times Europe print edition here.