If it would be possible for me to do so, I’d invite you, the reader, to travel back in time with me to a time long, long ago. Well, not THAT long ago – merely to the fall of 1945. It is, of course, a momentous time in history – not just because the Second World War has just come to an end courtesy the dropping of the two atom bombs on Hiroshima and Nagasaki (and we have acquired those reams and reams of black-and-white footage that will one day sustain the History Channel). The faculty and staff at the Moore School of Electrical Engineering at the University of Pennsylvania have assembled a grotesque monster of an electrical system. Its thirty separate units (plus power supply and forced-air cooling) weigh over thirty tons. Its 19,000 vacuum tubes, 1,500 relays, and hundreds of thousands of resistors, capacitors, and inductors consume almost 200 kilowatts of electrical power. It is of course a “Computing Machine” and has been dubbed the ENIAC. Of course, it does not take long for this new technology to get noticed. In 1950, the magazine “Popular Electronics” speculates “We have reason to believe that in the future, computers will weigh less than half a ton”.

Blip – we’re back to the spring of 2004. Today, a Palm Pilot that weighs less than 3 oz. probably has more computing power than the ENIAC. So how did we get here? We got here because we learnt to fabricate all those resistors, capacitors and transistors (that do the same work the vacuum tubes did) together as Integrated Circuits (ICs) on a chip. Today’s chips contain millions of transistors on every square inch of surface. We might as well pause to reflect that in doing so we have not only succeeded in doing the work of the ENIAC (scientific computation) quickly, cheaply and reliably on a much smaller platform, but have used the technology to create a range of products whose applications are so wide-ranging that it is safe to say that the creators of this technology had never even imagined them.

Time to switch tracks a bit. Let us try to imagine what an analytical chemist (or a bio-chemist or a molecular-biologist, or a medical diagnostic chemist) does today. If you’re like me – you’d imagine a person working at a lab-bench surrounded by test-tubes and other sundry lab-ware. She takes multiple samples from her source material (that can be something live), performs some sort of a separation process like filtering or centrifugation, mixes chemicals and waits around for a result (such as a color change) to show up. This would, indeed, not be too far from the truth. And now, if you’re an engineer, you’d realize that she used mechanical, optical, and maybe even electrical systems to perform a variety of operations on solids, liquids, suspensions and gasses. The devices or systems that she uses are like the big bad vacuum tubes of 1945. They are slow and require a comparatively large amount of material and energy. Moreover, they confine the user to a labor-intensive process. Now, if we could shrink the devices that she uses and integrate them together on to a single platform (like the computer chip), we could enable her to quickly find the answer to the question that she’s trying to answer (Such as – “Is this sample of blood HIV positive ?”). Such a device (a so called Lab-on-a-chip) might be called a Micro-Electro-Mechanical System (or MEMS) and would be to a lab (or a chemical factory) what the chip – which is in effect a micro-electrical system – was to the ENIAC.

Although he did not call it MEMS, the idea of a functional system in the micro (and nano) scales was first predicted way back in 1959 by Prof Richard Feynman – the Physics Nobel Laureate who is perhaps more famous for his unorthodox style and wonderful sense of humor than his work on Quantum Electro Dynamics. In a lecture titled “There’s Plenty of Room at the Bottom” he pointed out that making functional devices at the micro (and nano) scale was not physically impossible – and invited physicists and engineers to check this field out, saying “… it would have an enormous number of technical applications.”

Perhaps not too surprisingly, the first Micro Electro-Mechanical Systems were built by people in the IC fabrication community. The idea was to create “Integrated Sensors”. In a traditional macro-scale sensor, a mechanical element (such as a beam, cantilever or diaphragm) “senses” external stimuli (such as changes in temperature and pressure) by bending or deforming and this deformation is recorded as a change in electrical property (such as resistance or capacitance) by an electrical circuit that is connected to it. Integrated sensor technology sought to scale the mechanical elements down like their electrical counterparts, and put them on a single platform.

The technique adopted for doing so was also borrowed from IC manufacturing. At the heart of the IC manufacturing is a process called Photolithography. As the name suggests, it refers to creating structures using light. Typically a uniform coating of a low-viscosity resin (called photo-resist) is spread on a silicon wafer. This photo-resist has the unique property that is hardens when hit with UV light. If only a limited part of the layer on the wafer is exposed to UV light, the part that is exposed forms a solid crust but the rest remains fluid allowing it to washed out by an organic solvent like acetone. If the wafer is now dipped in a bath of etchant (such as hydrofluoric acid or potassium hydroxide), the areas on the wafer without the resin coating are selectively eaten away. By a clever use of multiple such steps, it becomes possible to fabricate mechanical structures such as diaphragms, beams and cantilevers.

Integrated pressure sensors got a mass market in the automotive industry in the late 70s and early 80s. The demand for improved fuel economy and stricter emissions standards persuaded car manufacturers to install systems that control the rate of combustion in the cars’ carburetors and these sensors played a vital part in that.

A more sophisticated example of such a device (which by now had begun to be called a MEMS device) was the accelerometer (the device that senses a rapid negative acceleration as a car crashes and provides the signal to detonate the air-bag). Other such devices which subsequently came into the market include gyroscopes, ink-jet printer heads, optical switches, micro-relays and sensor heads. These devices were seldom stand-alone products – with a few exceptions, they invariably interfaced directly with a macro-system to perform some task at the macro-scale. More importantly, they were each designed to fulfill a different requirement – and they just happened to be built using a spin-off from IC fabrication technology.

One manufacturing paradigm has been to build the components separately, and assemble them together on a platform to achieve the desired goals. This would potentially allow for economies of scale (Like micro-chips, it takes a lot of resources to set up the facility to make such devices – but once installed, a large number of them can be made relatively cheaply). While each individual MEMS product might have a limited market, if they all use the same components, the makers of the components (such as the pumps) would be able to make each piece much more cheaply and the cost savings would travel down to the consumer of the final products. This approach obviously relies on the assumption that assembling them will not be very difficult or expensive. Unfortunately, this has not always been true – and various solution have and are being explored to solve this fundamental “packaging” issue. One rather interesting technique, that perhaps has no macro-scale counterpart, is called Microfluidic Self Assembly. In this technique, we take two chemicals (say A and B) that have a strong affinity for each other. We coat the component with one – and the region on the platform where we want it to sit with the other. Now we just dip the platform (say a wafer) into a liquid bath which has the components suspended in it. The components just stick in the areas we want them to. This process can be repeated using different pairs of chemicals for different components of the system.

The other approach is of-course to fabricate the entire system at the same time. The platform of choice to achieve this goal may or may not be the silicon wafer. Certainly for a number of applications (such as bio-medical ones) Silicon is not the most suitable material. Neither is it necessarily the cheapest or the easiest to fabricate a given structure from. Its widespread use has primarily been due to the fact that the first MEMS were developed as spin-offs from IC manufacturing and silicon allowed easy integration to complex electronics

However, once people in other areas (including potential users of MEMS devices developed) got to know that “cool” stuff could be built in this way from Silicon, other materials began to come into the picture. For instance, ceramics (that have the ability to withstand high temperatures) have been used in certain micro-combustors and bio-compatible glass and polymers are being used for biomedical applications.

A critical requirement for almost any biological/biomedical device is “Biocompatibility” or the ability to remain in physical contact with living tissue or biological samples (such as blood) without causing an adverse reaction (such as an immune response). A large number of biological / biomedical micro-devices have been made out of PolyDi-oxyMethylSiloxane (PDMS or Silicon-rubber) using a technique similar to molding that is popularly referred to as soft-lithography. Another technique, called liquid phase photo-polymerization builds structures in a way similar to the original photolithography technique. However, instead of photo-resists, it uses the monomeric precursors to bio-compatible polymers, and one can create both rigid and flexible structures by adjusting both the amount of cross-linker mixed with the monomer and the duration of exposure. Thus we now have the ability to create almost any structure we desire.

There is, however, a special catch involved in designing MEMS or microsystems. Very often, one cannot merely take the design of a device that exists in the macro-scale, build a smaller version of it and expect it to function. This is because as systems get smaller, the amount of surface area per unit volume increases dramatically. [A cube whose edge is 1 m has a surface area of 6 m2. Thus its surface area per unit volume is 6 m2/m3. In contrast a cube 1 cm across has a surface area per unit volume of 6 cm2/cm3 which is equal to 600 m2/m3.] This changes the physics of the overall system – electrostatic forces, friction, surface tension and diffusion become major players whereas gravity and fluid turbulence often become negligible. The challenge lies in developing solutions that exploit the unique nature of these systems to achieve what we want to do.

Perhaps the area that is one of the most challenging and which also happens to have the biggest potential is the area commonly referred to as BioMEMS or Biomedical Microsystems. The impact of these systems is quite wide - from devices that manipulate tissue samples, to those that track and manipulate individual bio-molecules. A large number of companies are currently building lab-on-a-chip type platforms and systems for drug screening and for various types of genetic analysis. There are even some defense related applications of this type of technology. For instance, scientists at Sandia National Laboratories (Livermore, CA) are developing ChemLab® , a portable, handheld chemical-analysis system for homeland security, defense, and environmental and medical applications. ChemLab is being designed to detect chemical warfare agents and proteins, as well as various biotoxins. It should also be able to identify viruses and bacteria. Sandia currently has a prototype and expects to commercialize the system within the next two years.

It needs to be emphasized that like IC fabrication, MEMS is merely an enabling technology. Faster chips and larger memories have enabled the creation of many products and services that the original creators had no inkling about. Similarly, the next trillion dollar business idea may utilize MEMS but may not be just about MEMS. It merely remains to be seen what that big idea will be.

|