“What are the possibilities of small but movable machines? They may or may not be useful, but they surely would be fun to make.” —Richard P. Feynman, 1959
Like a tiny powerboat streaking across a shallow lake of grease, the tip of a scanning tunneling microscope skids through a film of engine lubricant. The slick coating parts behind the onrushing needle, exposing a vein of underlying platinum until the lubricant flows back in again. In the laboratory of physicist Jacqueline Krim, the needle’s mad dashes test a lubricant for possible use on microscopic machines.
The processes that produce microcircuits or chips can also crank out complex mechanical devices only micrometers across. A decade ago, micromachine enthusiasts and excited journalists painted visions of armies of self-propelled gadgets smaller than dust specks, serving mankind.
The pundits predicted minuscule mechanical disease fighters, for instance, that would roam our bloodstreams correcting the earliest consequences of illness. Stoking the excitement, a Japanese team built a tiny car, smaller than a grain of rice, from micromachined parts.
Indeed, a thriving industry has sprung up based on micromachines. As new products promise to break into big-money applications such as telecommunications and biomedicine, a gold-rush mentality has seized the field. “The market is really exploding. These are very exciting times,” says San Francisco-based micromachine market consultant Roger H. Grace.
The industry so far, however, is based almost entirely on micromachines such as automobile air bag accelerometers (SN: 9/26/98, p. 206: http://www.sciencenews.org/sn_arc98/9_26_98/Bob3.htm), whose moving parts, if any, are never intended to come into contact. For now, companies are content to produce such devices. More elaborate micromachines with parts that touch, made mainly in laboratories, are “great Tinkertoys,” says Jack Martin of Analog Devices in Cambridge, Mass. “But guess what. Nobody needs them.”
However, cars, robots, and most other sophisticated machines in the world today have moving parts that contact each other. Micromachines will someday “change the world,” predicts Krim, of North Carolina State University in Raleigh. But so far, they’ve failed to live up to their hype, she says.
The main reason is that early enthusiasts didn’t anticipate the powerful forces that arise at the surfaces of micromachines, Krim and other researchers say. Surfaces that come into contact tend to stick together because of molecular interactions known as van der Waals forces, electrostatic forces, and other influences. Microcomponents that both move and touch wear out quickly, if they can get moving at all.
“These surface issues have been problems, so people have gone for the low fruit,” says Kenneth S. Breuer of Brown University in Providence, R.I. “As all this microtechnology is exploding, these more complex physical problems become important.”
Lubricants such as those that Krim studies may help solve some of the problems. Researchers are also pursuing new coatings and alternative materials for making micromachines.
Yet many aspects of micromachine breakdowns still baffle investigators. “As people look at more complicated machines, the more subtle science issues will come back to haunt us,” says Breuer.
Micromachines’ contact surfaces are believed to span only tens of atoms on a side. Until recently, scientists hadn’t explored the behaviors of materials and surfaces at that scale. “The more experiments we do, the more we find things that we cannot explain,” says Stephen Hsu of the National Institute of Standards and Technology in Gaithersburg, Md.
To unlock the technology’s full potential, researchers must answer many fundamental questions. Moreover, Krim and other scientists say, micromachines are just a start. The next generation of even tinier, nanometer-scale machines is waiting in the wings. For it, the forces that have thwarted micromachines will have an even more powerful effect.
Veterans of the microelectronics industry started to make simple, prototype micromachines in the late 1970s. They dubbed the technology microelectromechanical systems, or MEMS, because a single microscopic device could incorporate both electric and mechanical components.
The development of electromechanical devices was “a natural evolution from semiconductor manufacturing,” says Michael T. Dugger of Sandia National Laboratories in Albuquerque.
To make microchip components—either electronic or mechanical—a fabricator deposits thin layers of materials such as silicon, silicon dioxide, and metals onto a substrate that’s usually silicon. The fabricator uses acids or ion beams to etch away sections of the films before applying the next layer. The building and etching sculpt the component. As the components of a micromachine are simultaneously created in this way, strategically positioned pads of silicon dioxide or other dispensable material enclose the parts of the machine intended eventually to be free. A final bath in hydrofluoric acid removes these pads.
In retrospect, the deeper reason for micromachine failures is obvious, Hsu says. As objects become smaller, their volume shrinks faster than their surface area. Forces such as gravity, related to volume or weight, become negligible, while forces such as adhesion and friction, related to surface area, take over.
Problems with adhesion—a failure mode dubbed stiction—often arise during fabrication. For example, the surface tension of fluids used in the final rinse can glue parts together. Stiction can also affect surfaces that accidentally bump into each other during manufacture.
Even when surfaces are free to glide past each other, the microscale mismatch between a micromachine’s strength and friction can be staggering. Bharat Bhushan of Ohio State University in Columbus is studying a micromachine built by French colleagues. Pushing the machine’s parts with an atomic-force microscope tip to get them to move, he and his coworkers have had to exert some 10 billion times the few piconewtons of force the machine itself can generate.
Micromachines wear out too soon partly because they’re usually made from silicon. It’s the ubiquitous material of the semiconductor electronics industry (SN: 3/25/00, p. 204: Looking for Mr. Goodoxide) but “not a very good material as a bearing or rubbing surface,” says Jorn Larsen-Basse, program director for surface engineering and materials design at the National Science Foundation in Arlington, Va. What’s more, silicon faces bristle with unsatisfied, or dangling, chemical bonds that readily hook up with nearby atoms, worsening stiction and friction, he says.
Some scientists are investigating ways to improve upon silicon or replace it. Sandia researchers reported last year that silicon microengines—basically a gear or gear train rotated by electrically driven rods—fail predominantly because of wear. Now, the team has developed a tungsten coating about 20 nanometers (nm) thick that dramatically reduces wear.
Instead of typically conking out after a few hundred thousand rotations, average lifetimes of the new microengines soared into the hundreds of millions of cycles, says Sandia’s Jeremy A. Walraven. The results are “very promising,” he says. Sandia scientists described the coating in April at the Institute of Electrical and Electronics Engineers International Reliability Physics Symposium in San Jose, Calif.
One extraordinary micromachine with rubbing surfaces is Texas Instruments’ Digital Micromirror Device. It packs more than a million movable mirrors on a chip. Inventor Larry J. Hornbeck, a physicist with TI Digital Imaging in Plano, Texas, defeated the usual enemies of MEMS. He used beams that flex instead of shafts that turn, aluminum instead of silicon, and an innovative anti-wear coating. Digital movie projectors, high-definition digital television, and other applications already use the chip.
For implantable biomedical MEMS, diamond may prove to be a micromachine’s best friend. Also at Sandia, John P. Sullivan, Thomas A. Friedmann, and their colleagues have found a way to relieve high internal tensions that arise when diamond is deposited as a film. These stresses typically cause the films to peel, distort, or even break their underlying silicon substrates (SN: 12/12/98, p. 383). The substrates “just shatter into little shards,” Sullivan says.
Heating the extremely hard, low-friction substance to 600ºC for a few minutes converts a small fraction of its carbon bonds from a tetrahedral arrangement to a flattened one, which relaxes internal stresses without ruining mechanical properties. “In the next year and a half, we’ll try to make complex gears, wheels, and hubs,” Sullivan says.
Efforts to develop diamond MEMS are also under way at Argonne (Ill.) National Laboratory. Some resourceful inventors have found ways to use high friction to their advantage. In the laboratory of Ming C. Wu at the University of California, Los Angeles, standard deposit-and-etch processing creates groups of hinged structures that initially lie flat on a chip. The prone structures hold minuscule mirrors, lenses, and other optical microcomponents. Beside them, fabricated in the same process, lie arrays of microrobots called scratch drive actuators.
Invented in the early 1990s by a Japanese researcher, scratch drive actuators use friction to propel themselves. In response to electric signals, each microrobot leans forward, flattens itself against the surface, and then drags its narrow foot a short distance, typically about 30 nm. Able to exert considerable force, the arrays of crawling robots push and pull on the flattened structures until they rise up like decorations on a pop-up greeting card.
In June at the Solid-State Sensor and Actuator Workshop on Hilton Head Island, S.C., Pamela R. Patterson, Wu, and their coworkers described a lightweight, tiltable micromirror that scratch actuators raise into position hundreds of micrometers above a chip surface. It can then pivot through large angles to direct laser beams.
Although MEMS makers have learned how to avoid stiction in some devices and to design away the worst friction and wear in others, they say they’re groping in the dark for answers to basic questions. What’s happened, explains Krim, is that micromachines have opened up an unexplored realm of fundamental surface physics.
The technology “has crashed head-on with science,” Krim says. “Friction and lubrication at the atomic scale are totally different than at the macro scale. You can’t squirt WD-40 on [micromachines], it will lock them shut” by the surface tension of the liquid, she says.
Krim is concentrating on lubrication questions. One of the basic goals of her racing-needle study is “to look at why a lubricant lubricates,” she said at the March meeting in Minneapolis of the American Physical Society.
In tests with a needle moving at high speeds, one lubricant that seems promising is tertiary butyl phenyl phosphate, or TBPP, an additive for automobile engine oils, she reports. Instead of wearing out, the lubricant “becomes more slippery as conditions become more brutal,” she notes. It may be chemically reacting with the surface it’s coating, perhaps instigated by heat from friction, she speculates.
Scientists know little about such “tribochemical” reactions or the conditions that spark them, Krim says. Besides slashing friction, the reaction may also be causing corrosion or other damage.
As part of a military-funded project to make a jet engine on a chip, Breuer is investigating the use of air or other gases to lubricate extremely high-speed MEMS. At Hilton Head in June, he and his colleagues described prototype micromachine turbines that spin at up to 1.4 million revolutions per minute.
Although often not directly motivated by concerns about MEMS technology, fundamental studies also promise to help solve some of the puzzles. Theorist Mark O. Robbins of Johns Hopkins University in Baltimore and his colleagues have been exploring the atomic-level origins of static friction—the force that must be overcome to set an object sliding after it’s been resting on a surface.
Curiously, calculations on crystal surfaces by several researchers have indicated that, except in the rare cases when lattices match perfectly—both in spacing and alignment—there should be no static friction. However, something else must be going on, Robbins notes, since objects “would then slip off everything.”
According to the Amontons’ laws, recognized 300 years ago, the force of friction grows in proportion to how strongly surfaces are pushed together, and its strength is independent of the area of contact. Last year, Robbins’ team unveiled computer simulations indicating that small molecules of everyday grime can lock surfaces together, producing static friction. “The dirt on all surfaces naturally gives this kind of law,” he says.
In April, Robbins and his coworkers posted an extension of those findings on the online physics archive (http://xxx.lanl.gov/abs/cond-mat/0004494). The scientists found that their molecular-scale dirt may cause not only static friction but also kinetic, or sliding, friction.
What’s more, the group showed that the findings apply not only to crystalline surfaces but also to amorphous ones, the kind common in micromachines.
When will the MEMS train finally leave the station? “The distant future of MEMS looks very promising, but it’s distant,” says Bhushan. A more optimistic Dugger predicts that “new applications involving contacting and rubbing surfaces are on the horizon.”
Krim compares today’s micromachine to the automobile—circa 1916. Referring to a historical account of car maintenance from that year, she notes that keeping a new Maxwell running meant wielding an oil can many times a day plus emptying tube after tube of grease into its parts every week or month.
Since then, “the automobile has come a long way,” she observes. “I believe MEMS devices will, too.”