Physics and Technology Forefronts

Micro Electro Mechanical Systems: Pyrite Or Pure Gold?

by Eric Peeters, Xerox Palo Alto Research Center

Imagine a machine so small that it is imperceptible to the human eye. Imagine working machines with gears no bigger than a grain of pollen. Imagine these machines being batch fabricated tens of thousands at a time, at a cost of only a few pennies each.Imagine a realm where the world of design is turned upside down and the seemingly impossible suddenly becomes easy.. Welcome to the microdomain, a world now occupied by an explosive new technology known as MEMS.MEMS is the next logical step in the silicon revolution.We believe that the next step in the silicon revolution will be different, and more important than simply packing more transistors onto the silicon."
[excerpt from http://www.mdl.sandia.gov/Micromachine/vision.html]

".MEMS'97: Show me the Money !.",
".Widescale progress but still no killer applications.",
".MEMS market data: Another case of sorry-wrong number ?."
[excerpts from 'Micromachine Devices' (from the editors of R&D Magazine), issues March and August 97]

Superposition of a false color 129Xe image onto a 1H spin-echo image showing that the 129Xe signal arises entirely from within the brain as well as reduced 129Xe in the cerebellum. Image from http://bearhat.rad.med.umich.edu/˜scott/xeimages.cfm

These quotes from within the community express the existing spectrum of sentiments about the significance of a technology called Micro Electro Mechanical Systems (MEMS, US), aka Micro System Technology (MST, Europe), aka Micromachines (Asia) and about its impact on society two decades after the birth of the field. In 1982, Professor Kurt Petersen of Stanford University helped launch a new field by authoring a truly visionary publication, Silicon as a Mechanical Material,¹ in which he advocates the notion of using micro-electronics processing techniques and microelectronics materials to build microscopic parts with a mechanical function, in addition to the usual transistors and other electrical components. In 20 years, MEMS have literally gone from the Proceedings of IEEE to airline in-flight magazines, and MEMS technologists have amazed the world with awe-inspiring technological marvels constructed on micro-mechanical chips with intricate moving parts. But while anyone must admit the scientific and technological achievements have been magnificent by any measure, the opinions about how and when the long anticipated widespread economic impact might materialize are much more varied.

To the scientist or technologist, MEMS is like a dream come true. There is something magical about flicking on the electron microscope, zooming in and wandering through a micro-mechanical landscape that you thought up, created and understand, a world where laws of nature don't behave as the layman or novice expects based on their intuitions cultivated in the macroscopic world. The micro-machinist's building site is the size of a grain of rice, the architecture the size of a human hair, the building elements smaller than a red blood cell, .but our lithographic backhoes are the size of a city so we can construct a thousand sites at once. We have mastered the art of building the impossible, from gyroscopes, micro-motors, gear trains and transmissions, to fluid pumps, x-y tables and entire self-assembling optical bench erector sets.

The looks of this world may be wondrous, but its behavior is even more so. Despite our understanding of how bulk forces are supposed to scale down more quickly than surface forces with decreasing length scales, and despite our understanding that gravity effects should therefore disappear at some point, seeing something like Manhattan sitting comfortably balanced on top of something like the Empire State building takes some getting used to. (see photo caption...) especially if Manhattan is being moved back and forth electrostatically at a rate of 10 KHz with motion control much better than a city block. Your safest bet when first venturing into the world of MEMS, is to throw life-long intuition out the door altogether and to start cultivating a whole new intuition based on new experiences in the micro world.

Enamored by this wondrous new frontier and its parallels with microelectronics, we technologists have declared MEMS to be the second semiconductor revolution. We have claimed unconditional applicability of Moore's law and economies of scale, we have been predicting economic hockey stick curves and many of us believe that MEMS will soon become as pervasive in all aspects of every day life as microprocessors are today.

Three factors drive Moore's law in microelectronics: 'smaller is better,' 'smaller is cheaper' and the 'building blocks are universal across applications.'

However, none of these drivers is particularly valid for MEMS. The third is especially problematic. At the highest level of abstraction, the real power of microelectronics is not even its massively parallel fabrication paradigm. It is the existence of a generic element, such as the transistor, which allows us to build extremely diverse functionality 'simply' by implementing appropriate interconnection patterns within large collections of the generic elements. This is what makes semiconductor economics so vastly different from anything seen before in history. The impact of pushing the generic components along Moore's curve is therefore universal across all imaginable application areas, which in turn justifies massive spending on pushing even further along the curve. The 'gain factor' in the financial feedback loop is greater than one because of the 'generic element' paradigm.

In contrast, MEMS, by its very nature, does not have a set of generic elements. There is no MEMS 'transistor'. MEMS 'touch' and 'participate' in the physical world of the mixed bag of applications and therefore need to be much more application specific and less generic. In every aspect of design, modeling, manufacturing, packaging, etc. Thus, it is much more challenging to keep the gain factor that drives Moore's law greater than one. This is where many of the economic parallels with microelectronics break down and economics is evidently what makes the difference between a possible future and a likely future for a technology.

Although a 'second' silicon revolution with a magnitude close to the 'first' seems like a very tall order indeed, let there be no doubt that numerous MEMS-based products will enrich all of our lives in years to come. The economic successes may come via three routes:

The 'killer' applications may justify investments in totally dedicated manufacturing facilities. The revenue stream required to justify this sort of endeavor points to high volume products that are enabling at the function/system level itself and that are preferably accompanied by the sale of high margin consumables or renewable services. Inkjet printheads are an existing example of this route to success.²

MEMS technology that is buried deeper, at the lower component levels of the 'food chain' and not intimately coupled to a secondary high margin revenue stream —j such as from a consumable or renewable service — will be under pressure from the increasingly vigorous 'race to zero' in hardware cost. It may be that this class of MEMS components can be sustained over time if 'piggybacked' on existing or moderately customized semiconductor infrastructure. The ADXL series airbag accelerometers from Analog Devices have been an example of this route to success.

The third route is through the 'MEMS foundry' model, which may be viable in the long run if conceived with a fairly sharp application focus. If successful, this route may end up accounting for the majority of MEMS success stories in the next millennium.

Gold or Pyrite? No doubt there are plenty of pure gold nuggets in the MEMS ore. We have found some already. The next ones may well be in low-power reflective direct view displays for Personal Digital Assistants (PDA), or in wireless communication components that will allow your PDA to have access to the Web anywhere, anytime, and make it a pocket sized infinite information source. They may be found in the optical micro-mirror matrices that will switch broadband fiber communication to your home, or in disposable DNA diagnostics chips that won't leave you worried for days about the results of a critical blood test, or in.

¹K. Petersen, Silicon as a Mechanical Material, Proc. IEEE, Vol.70, No.5, 1982, pp.420-457.
²E. Peeters, Challenges in Commercializing MEMS, IEEE Comp. Sci. & Engr., Jan-Mar 1997.