MEMS are defined at the micron level by VLSI circuit processing techniques, giving technologists precise control over the finished product. Nanotechnology, which uses the molecular-scale processes of chemistry and living cells, is based on harnessing their molecular interactions to set in motion processes that create some desired end configuration.

One recent result in this direction was a hybrid micromachine/DNA system, announced by IBM researchers at the company's Zurich research center. In the device, tiny micromachined cantilevers were selectively deflected by DNA fragments. The prototype demonstrates for the first time a machine that is driven by DNA molecules. The work, done in tandem with a group at the University of Basil, was reported in a recent issue of the journal Science.

DNA power

The device uses the lock-and-key mechanism of DNA chemistry. An array of cantilevers is treated with different strands of DNA. When a solution containing different fragments of DNA is introduced, complementary strands of DNA will naturally bind to specific cantilevers. The bonding process creates stress which deflects the cantilever. The effect has been applied so far to detecting damaged DNA sequences, since a single base mismatch will cause a slightly different stress, indicating the presence of a damaged fragment.

"The ability to use biology to perform specific mechanical tasks on the nanometer scale with silicon provides a completely new approach to operate machinery autonomously, without external power or computer control. We have found a way to get DNA to do the work for us, so we don't need batteries, motors or the like to operate tiny machines," said James Gimzewski of IBM Research, in a statement released by the company.

This demonstration complements work being done at Zyvex LLC (Richardson, Texas) in which molecular-scale systems are being structured a molecule at a time in tiny micromachined manipulators. Founder Jim Von Ehr said that real applications are at least 10 years away. At Zyvex, the professed corporate goal is to build a zybot nanotechnology assembler — a device capable of assembling macroscopic substances from microscopic materials with atomic precision. Such a zybot builds tiny manufacturing plants capable of creating bulk materials and nanoscale structures with atomic precision by putting every atom in a blueprinted location. Zyvex's nanomanufacturing capability is enabled by a new technology called mechanochemistry — that is, atomic-site-specific chemical reactions resulting from precise positional control of the reactants in combination with mechanical or electrical force applied to overcome the reactions' quantum energy barriers.

Zyvex has built a molecular assembler, which employs a limited number of molecular building blocks to assemble larger more complex molecules. This first prototype will not be able to build arbitrary materials. It will likely use only a single microscopic assembly "tool" under the control of a CAD program.

"The nanotechnology assembler will expand on the molecular assembler not only by being smaller but by employing many identical tools that self-assemble by replicating themselves into a predetermined number of parallel tools, all of which can be controlled from a single CAD program," said Ehr.

A prototype nanomanipulator has been built to perform operations in the vacuum chamber of an electron microscope. Using tiny micromachine actuators driving atomic-force microscope (AFM) tips, the tool can build objects out of carbon nanotubes. One of the first tasks of the manipulator will be building better AFM tips. A group at Washington University (St. Louis, Mo.) is studying the possible operations that can be performed with the machine.

Such micron-scale machine tools, using probe techniques borrowed from microscopy, may become the enabling technology for building nanoscale systems. For example, the machines could be employed to build specific complex molecules that could be the basic building blocks of what Ehr defines as mechanochemistry. This would be a new variant of chemical manufacturing in which specifically tailored molecules are positioned within a reaction and mechanical or electrical force is applied to drive constituents of the reaction together. Conventional chemistry relies on the diffusion of molecules through the reaction, as the principal mechanical component that drives chemical synthesis.

Living cells, while employing chemical reactions, are in some ways closer to mechanochemistry in that really large molecules like DNA, RNA or proteins drive the reaction. However, as Ehr pointed out, these reactions take place close to thermodynamic equilibrium, which limits their range of action. Mechanochemistry operates far from equilibrium, producing fast, high-energy reactions that could forge fundamentally new types of molecular machines. The scope of mechanochemistry is identical to that of nanosystems, as defined by Eric Drexler, who popularized the concept in his book Nanosystems: Molecular Machinery, Manufacturing and Computation, which was published in 1992. The term was first proposed by quantum physicist Richard Feynman at CalTech in 1959 in a now-famous lecture titled "There is Plenty of Room at the Bottom." Feynman's proposal remained a subject of speculation until recently, as scientists began to manipulate individual atoms with atomic probe microscopes and the fine detail of VLSI chips continues to shrink at geometric rates.

Techno revolution

Drexler is motivated in part by what he sees as the need to seize the initiative in what could be a profound technological revolution. In addition, government funding bodies are beginning to view the concept with some seriousness. The benefits of fabricating nanoscale devices with atomic precision prompted Japan's government to launch a $200 million program called the Ultimate Manipulation of Atoms in 1993-less than a year after Drexler's book. In the United States the National Science Foundation has supported work in nanofabrication as has many government funding agencies.

Still, Drexler maintains that other countries are ahead of the United States "The world is already a nanomachine," he said. "U.S. scientists have just been slow to wake up to the fact. Look around you. Every cell in every plant and animal is a nanomachine. Each cell has about a billion bytes of blueprints plus the nanomachinery with which to manufacture it." The human body has 100 trillion of these billion-byte cells.

Despite the successes of current MEMS, the smallest mechanical manufacturing technology to produce commercial products such as accelerometers and micromirror arrays, not a single nanosized device that performs a useful function has yet appeared outside a lab. The greatest promises of nanotechnology mavens have so far prompted only more research. In defense of the concept however, Drexler points out that a lot of infrastructure still needs to be invented before even a nanotechnology foundry can benefit working engineers. "Making such small devices involves much more than just the computing element," said Drexler. "There is also the wires, mechanical stuff, interfaces and other things which every engineer needs in order to build working systems."

It is the factor of manufacturing infrastructure, not size, that distinguishes MEMS technology from nanotechnology proposals. MEMS have been able to spin off from an established process, so an engineer can take a design to a foundry today and get working parts within spec and on time.

Unfortunately, the existing process is essentially planar, since that's where circuit manufacturing started. That puts restrictions on the mechanical components that can be built. MEMS researchers are starting to tackle that problem by devising multilayer processes. For example, at Sandia National Laboratories (Albuquerque, N.M.) research has evolved a five-layer process that the lab expects to license to a foundry this year.

Tool development

Meanwhile, nanotechnologists are starting to tackle the design and simulation tools that any nanotechnology process will require. Drexler said that the Institute for Molecular Manufacturing (Palo Alto, Calif.), where he is a research fellow, has simulations that can design and simulate configurable matter. As a start, Drexler cited the Institute's molecular designing techniques using CAD software to give a better definition of just what configurable matter might look like. Drexler said chemists have learned from nature how to make long chains of molecules like proteins but, for configurable matter, small bricks that stick together neatly and precisely are what's needed.

"Proteins with the right sequences can fold up to make a small compact object involving multiple chemical and molecular factors. Some molecules like to be on the surface, and some inside — what you want built into a CAD system is software that can design the right molecular building blocks and put them together with a well-understood searching algorithm to find the right materials," said Drexler. Once designs are on the drawing board, tiny MEMS-based assemblers could presumably begin to realize actual nanoscale machines.

Many of the basic principles of nanotechnology are already clear. For instance these nanosized devices are just too small and numerous to be assembled manually, so self-assembly is necessary. For Drexler, the goal is to make self-assembling tools that in turn can be used to make non-self-assembling nanomachines that then build the required nanoscale devices. Regardless of the exact path to the same goal, a range of products will likely astound the unsuspecting with specs like a supercomputer that fits on the head of a pin. Another possibility would be a mechanical computer using 1-nanometer-diameter diamond rods that execute a billion instructions/second in less space than a cubic micron while consuming only 100 nanowatts.

Engineers will be able to transcend the laws of nature — or appear to — Drexler said, by designing different actuators on different sides of nanosized building blocks. That would make it possible to create materials with seemingly impossible properties, such as a material that never flexes all the way to its breaking strength. On a microscopic level, the real flexing could be compensated for by extending different actuators on different parts of the material's building blocks, providing the appearance of zero flex at the macroscopic level.

"Smart materials will have remarkable properties, such as being stronger than steel but more transparent than glass and with the density of balsa wood. We will build aeronautical materials that are 70 times better than the current best strength-to-weight ratio," said Drexler, who sees a direct link between nanoscale systems and VLSI technology. "When I wanted to build my first nanoscale system I just picked up a copy of Introduction to VLSI Systems by Carver Mead and Lynn Conway and found that it was remarkably applicable."