The Road to MEMS
May 24, 2012, MEMS Business Forum, Santa Clara, CA—Kurt Petersen from KP_MEMS gave a historical perspective on MEMS. The ability to integrate complex functions into a chip has led to high volumes and lower costs.
Starting in '75, there were pressure sensors, strain gauges, and thermal print heads. Now, the range of applications using MEMS spans consumer electronics, microphones, oscillators, digital light projectors, variable optical attenuators, IR image detectors, gas chromatographs, microfluidics, and medical devices.
The devices are now shipping in the billions of units a year, and have grown from less than 1 percent of the semiconductor market in '85 to over 3.5 percent today. The technologies have reduced the size of inertial measurement units from a very large bread box to a single IC package. Sensors have moved from expensive, low-volume military subsystems to high-volume consumer devices.
The drivers for this size reduction are correlation with function and the on-going scaling of semiconductor processes. If a function has no correlation with its size, it will shrink. Electronics is a perfect example of correlation with size and function. The continuing reduction of device features means that we can put more functions into the same size or smaller package.
The semiconductor industry has been working on improving and reducing feature sizes in their technologies. MEMS, however, has been a stepchild with hand-me-down equipment and a lot of custom specialty production equipment. Now, fortunately, the equipment suppliers are working hard to design and produce this specialty equipment because the market is growing faster than standard semiconductors.
The economics and competition are also contributors to the MEMS growth. When markets have products based on difficult-to-use technologies, the products are ripe for displacement with MEMS. For example, driven wire dot matrix printers were replaces by ink jet printers. Gyroscopes changed from expensive and delicate quartz assemblies to MEMS, electret microphones could not be soldered into boards, quartz oscillators have a 50x higher zero-hour failure rate than MEMS.
These and many other examples show why some MEMS functions are seeing such rapid adoption. As sensors and actuators moved from military to industrial to consumer products, the benefits of size reductions and their accompanying cost reductions have opened the markets for alternative sensor types.
The MEMS industry owes much to the major universities who led the basic research into processes and structures, and many early device demonstrations. These institutions continue to contribute through on-going research and a continuing supply of new MEMS students. The NSF and DARPA have been finding many of these projects for over 25 years. This is not to leave out the extensive corporate R & D investments for MEMS.
The technologies for manufacturing and production require specialized equipment and chemicals. The materials processing industry has responded with custom and standard processing equipment for the specialized needs of the MEMS community. Even the packaging companies have created new packages to accommodate the special needs of MEMS devices. Some of the first package on chip and package in package devices were made for MEMS devices.
So will MEMS become a $1T market? To get there will take a 100x increase in sales from today. It took 35 years for a 100x increase in sales for the semiconductor market. Where will the units come from? Not medical monitoring at only 7B units. Wireless infrastructure and environmental monitors could get us there, but the support technologies will take years to be put in place. A guess is 25 years to get to 1 trillion units.
Currently, MEMS is where semiconductors were when the PC was introduced. MEMS have grown from 1 percent of the market in '81 to over 3.5 percent today. MEMS revenues are growing at about 15 percent a year, much faster than the overall semiconductor market. The infrastructure is in place to fuel growth.