Science / Medicine : ‘Microbots’ on Horizon : Technology Is Thinking Small
Microcircuits, fabricated on silicon chips, have produced a revolution in electronics. Now researchers are using the same chip-fabrication techniques to build micromachines, complete with motors, gears and other moving parts--all smaller than a dust speck.
And the combination of electronic circuits and micromachines, all on the same chip, offers the prospect of microscopic robots.
“Clearly this will be a new industry,” said George Hazelrigg of the National Science Foundation, which is funding some of the basic research. “I think it’s going to be a big one.”
This field is so new that it lacks even an agreed-upon name. At AT&T; Bell Laboratories, the term is “microdynamics.” Japanese researchers speak of “micromachines.” Rodney Brooks of the Massachusetts Institute of Technology talks of “microbots” and “microbotics.” Others use terms such as “micro-electromechanical devices,” a lack of settled terminology that illustrates the unsettled nature of the field.
Indeed, its beginnings trace only to the late 1970s. By then, techniques were well in hand for etching silicon to make microcircuits. Various small groups were using these same methods to make specialized instruments. A tiny pit etched in silicon, roofed over with a thin and flexible layer, could offer a microscopic pressure sensor, with the thin layer flexing under pressure.
In 1982 Kurt Petersen, then an IBM researcher, wrote a pathfinding paper that established a firm foundation for this field. “There were little sporadic efforts, people working on things all over the country,” he recalls. “They didn’t know each other. They hadn’t really grasped the idea that they were making mechanical devices out of silicon, which is an electronic material.”
Since then, the development of silicon micro-instruments has flourished. In late-model cars, a jewel-like pressure sensor, no more than two or three millimeters in size, rides beneath the hood. Auto makers buy them by the millions as part of the system that controls the air-fuel ratio in engines.
Also, the Pentagon’s Strategic Defense Initiative is developing miniaturized guidance systems featuring accelerometers the size of a match head. They feature a roofed-over pit etched in silicon, but the thin roof is cut away at the edges, allowing it to flex like a diving board when its rocket accelerates.
“We couldn’t form them by machining a block of metal,” said Brig. Gen. Malcolm O’Neill of the SDI. “Small, finely machined parts are hard to make.” But silicon etching methods can be controlled to far greater accuracy.
Hazelrigg expects that within five years, the next generation of microdevices will be ready. These will feature cutting wheels and other tools with moving parts, driven by motors the width of a human hair. These would not be robots, since they would not move or make decisions on their own. Instead they would be particularly precise surgical tools manipulated by a doctor, or highly complex and compact mechanical systems that would be under the control of an outside computer.
The necessary motors and wheels are under vigorous research. At Bell Laboratories, William Trimmer and Kaigham Gabriel have fabricated sets of three meshing gears, with diameters as small as 0.005 inches--about a hair and a half. A similar wheel, powered by air from a hypodermic needle, spins at 24,000 revolutions per minute. “As with microchips, we get hundreds of thousands of devices off a single wafer of silicon,” Gabriel said. “If one turbine breaks, we just reach in and get another one.”
To build a micro-electric motor, the wheel must be made to spin using electricity rather than air. At UC Berkeley in 1988, a group of graduate students led by Richard S. Muller built a micro-motor whose rotor, or turning wheel, was only two-thirds of a hair in diameter.
The researchers arranged a microcircuit that would apply charges of static electricity around the rotor’s periphery, shifting the charges’ locations rapidly. The rotor turned by following the electrical attraction of the charges. Muller declared that “the important thing is that now we know a rotor this small can work. We had no proof before.”
A more recent advance, at the University of Utah, addresses a major problem with such motors: They are of little use when spinning at very rapid speeds. Instead, they must turn a shaft at much slower speeds but without wasting power. The shaft then has torque, or turning force, and can do useful work.
The Utah invention is the “wobble motor.” Its shaft rolls around the inside a cylinder, leaving only a small gap between rod and cylinder. Hence the rod must make 40 or more “wobbles,” or rolls, for each rotation it makes, a design that paradoxically minimizes friction.
Steven Jacobsen, the project leader, declares that the longest-running microscopic motor, prior to his wobble motor, ran for only three minutes before stopping. His motor, by contrast, has operated continuously for a month.
For now, all such mechanisms reside at the level of basic research. “We don’t know anything about anything,” Hazelrigg said. “We don’t understand friction. We don’t understand mechanical properties. Under certain circumstances, smooth parts stick together easily,” rather than sliding freely. “We are exploring these issues with fundamental research.”
A main source of funding is from Hazelrigg’s NSF, at $1.5 million a year. AT&T; and a few other companies generate their own research funds. The Japanese are also involved and are reporting their research in a new journal, “Micro Machines.” The University of Tokyo has arranged for Bell Laboratories’ Gabriel to come for an extended visit. Also, Toyota is hosting a workshop in Japan this October and is so eager to learn from American specialists that the firm is paying their travel and expenses.
Hazelrigg notes that it costs about $400 to fabricate a silicon wafer, whether of micro-electronics or micromechanisms. Each wafer yields about a hundred chips, and each chip could have up to 100,000 micromotors or similar devices. Each motor, then, would cost less than 0.01 cent, allowing them to be assembled--and discarded--with abandon.
The University of Utah’s Jacobsen hopes to use such arrays of micromotors, covered in natural-looking synthetic skin, to build an artificial hand for amputees. It would need large numbers of standard motors, each acting like an individual muscle fiber, to give the desired dexterity.
Kurt Petersen expects to use micromachines to accomplish the highly tedious task of attaching tiny wires to microchips of present-day design. This task is ideally suited for robots, but today’s versions lack the needed dexterity. Instead, workers in assembly plants in South America or Asia squint through microscopes to manipulate the wires.
“We damage their eyesight,” Petersen added. “The young girls come and they have a lifetime of three to four years before their eyesight deteriorates.” But micromachines would both improve working conditions and eliminate an industrial hazard, allowing manual workers to become machine operators.
Hazelrigg expects that micromachines will be used in surgery on the eye, brain or heart. “The surgeon would pass such a device into the body through a hypodermic needle,” he said. “The trauma of opening the chest will be replaced by that of an injection--it’s nothing” The surgeon would control the microtool with a hairlike rod passed through the needle, cutting tissue virtually cell by cell.
Researchers expect micromachines will evolve into true microbots, able to move entirely on their own, powered with solar cells or rechargeable batteries. The next generation may well have on-board sensors and computers, which would give them eyes and brains.
“Dumb” versions would have only limited abilities but could be highly useful.
Anita Flynn, a researcher at MIT who is designing microbots, expects that a “gnat robot” could repair a break in an electrical line. It would feel its way along the wire, checking for the break. When it found it, the robot would simply lie down in place, its body serving as the needed connector.
Her colleague Rodney Brooks expects to build a flying version, the size of a miniature model plane. It would fly over farmers’ fields, using an infrared sensor to avoid obstacles, to check on the crops. If the sensor found the plants needed water, the plane would soar down and signal an electronically controlled irrigation valve.
Household microbots might spill from a vacuum cleaner, then head for the corners of the room where dust collects. Using moving combs, they would busily rake dirt particles from the carpet. Then, at a signal from the vacuum cleaner, they would return to the center of the room and spill their contents in a pile, which the vacuum cleaner would scoop up.
Other microbots would be smarter, being equipped with better imaging systems. Brooks envisions “agrobots” as a replacement for herbicides and insecticides. They would be able to tell good from bad plants and would cut down the weeds with a tiny saw. The agrobots would also be on the lookout for any small object that moves. “It’s probably a bug, so you shoot at it,” Brooks said. “Maybe with a javelin.”
“There are endless possibilities,” he added.
MICRO-FABRICATION TECHNIQUE
To fabricate a silicon accelerometer one exposes the silicon to steam, which produces a thin layer of glass, silicon dioxide, upon the surface. This surface is protected by a chemical photoresist, liquid plastic that breaks down when exposed to light.
The light shines through a finely-detailed photographic mask to react with the photoresist in a pattern similar to the mask. Strong chemical solvents are then used to remove the unprotected areas of the glass layer.
Other etchants are used to remove areas of the silicon that are no longer protected by the glass layer. With this process, it is possible to fabricate an accelerometer as a long thin beam that overhangs a cavity or pit.
The pit and beam together form a variable capacitor, an electrical device whose ability to hold electric charge varies with the beam’s position. In this fashion, the accelerometer can turn a mechanical deceleration into an electronic signal. These will be used in crash sensors, to trigger the release of air bags. A true collision--as opposed to the shock of hitting a parking lot post--has a characteristic ‘signature’ featuring a deceleration that is both intense and relatively prolonged. The sensor would use electronics, linked to the accelerometer, to recognize this signature.
