Gulliver would be right at home. By etching microscopic components into the surface of silicon wafers, researchers are building Lilliput-sized engines that work just like their full-sized brethren. Using various techniques pioneered by micro-electro-mechanical systems (MEMS), these penny-sized engines can power miniature pilotless aircraft and satellites. Coupled with miniature generators made the same way, microengines could replace batteries in sensors, laptops, or any other device needing portable power. And unlike conventional batteries, which take time to recharge, generators-on-a-chip could be refuelled in seconds—with little more than a squirt of lighter fuel.
One of the most enthusiastic supporters of microengines is the Defence Advanced Research Projects Agency (DARPA) in Washington, DC. William Tang, the agency's programme manager for MEMS research, believes that a power source built into a silicon chip would be extremely useful on the battlefield. One device being studied is a sensor no larger than a sugar cube that contains all the wireless communications and data storage circuitry along with the power supply. A microengine generating up to 50 milliwatts would be enough to power such a sensor as it collected information from its surroundings and beamed the data up to a low-flying aircraft. Hundreds of sugar-cube sensors could be scattered over the terrain and left to monitor ground vibrations and temperature changes for months on end.
More attractive still would be microengines capable of generating power in the 10-20 watt range. At present, soldiers have to carry 20-pound backpacks of batteries on the battlefield to feed their power-hungry communications and targeting gear. With further development, Dr Tang believes that today's 20-pound battery packs could eventually be replaced by a MEMS engine and its butane fuel can weighing no more than a pound.
Designing a miniature engine demands more ingenuity than merely shrinking the size of the components. Where gas flows are involved—as they are in the intakes, combustion chambers and exhausts of piston engines, gas turbines and rocket motors—the flow pattern has to be scaled along with the dimensions. That means the Reynolds number (a critical parameter that reflects a fluid's density, velocity and viscosity as well as the dimensions of the object it flows around) has to be the same in a microengine as it is in a conventional motor. Modelling the internal flow patterns on a much reduced scale is a highly non-linear business. For instance, if the dimensions are shrunk 100-fold, the flow speeds may have to increase 1,000-fold or more.
Jet engine on a chip
This is the problem Alan Epstein has had to solve at Massachusetts Institute of Technology, where he and his team are building a miniaturised gas turbine. Although Dr Epstein has plenty of experience designing conventional gas turbines, his group had never encountered the constraints imposed by microscale manufacturing before. If it is to achieve the same flow pattern, the rotor in a MEMS turbine has to spin a good deal faster than that in a conventional jet engine.
The fuel combustion process has also posed a challenge for the designers. Chemical reactions take place at the same rate whether they are occurring over one metre or one micrometre. As fuel flows through a microscale combuster, it therefore has less time to react than if it were flowing through an ordinary combustion chamber. Complicating matters further, the burning fuel cannot be allowed to get too volatile for fear of cracking the microscopic engine's casing, which is made of silicon instead of high-temperature alloy.
Earlier this year, Dr Epstein demonstrated that a prototype of his microscale jet engine could achieve rotor speeds of 1.3m revolutions per minute—enough, in theory, to generate 17 watts of power. The prototype is carved out of six wafers of silicon, stacked on top of each other like a pile of pancakes. The entire device contains just one moving part: the spinning turbine disc. Dr Epstein reckons that, fuelled with hydrogen, a MEMS engine measuring two centimetres wide and three millimetres thick could generate ten watts of electrical power. If a hydrocarbon fuel were used instead, it could supply several times more. The best batteries around today can deliver only a thirtieth the power from such a package.
A wholly different set of challenges has had to be overcome by Wei Yang at Honeywell in Morristown, New Jersey. Dr Yang is working on a microengine that uses a free-piston design, where the combustion process drives a piston which, in turn, drives a turbine. So far, the Honeywell team has two important achievements to its credit. First, it has managed to get fuel to ignite in volumes too small for a spark or fuel injection to work. By compressing the fuel extremely rapidly, Dr Yang has been able to make it burn in a cavity as small as one cubic millimetre. This removes the need for creating a special chamber for combustion.
The second breakthrough has been to get the combustion process to take place under extremely lean conditions. Traditionally, internal-combustion engines have used an air/fuel ratio of eight or more to one, and converted chemical energy into mechanical energy with an efficiency of 20-30%. By contrast, Dr Yang's device uses an air/fuel ratio of around 40 to one—even higher than the lean burn conditions found in the "stratified charge" engines of modern cars. As a result, the Honeywell device achieves an efficiency of over 50%.
Such an engine would be light and powerful enough to propel a "micro-air-vehicle"—a propeller-driven aircraft with a wingspan of 15 centimetres or less. Alternatively, by arranging the piston to move through a series of magnetic coils, its reciprocating action could be used to generate an electric current instead of mechanical output. When complete, Dr Yang expects his engine to be no more than one cubic centimetre in volume, and deliver five to ten watts of power.
There is no escaping the fact that, ultimately, micromachines will have to have microengines to drive them
But before any of this can happen, numerous engineering problems have to be solved. For instance, though internal-combustion microengines have high power-to-weight ratios, they are as bad as their big brothers when it comes to noise, vibration and noxious emissions. That may not matter to soldiers on the battlefield. But mobile-phone users could not tolerate them. Microengines are therefore expected to find their way into applications such as satellite propulsion systems before invading the laptop and mobile-phone markets.
Meanwhile, David Lewis of TRW, an engineering firm in San Diego, has pioneered the use of MEMS in rocket propulsion with partners at the California Institute of Technology in Pasadena and the Aerospace Corporation of El Segundo, California. The group has built an array of microthrusters from three strata of glass and silicon. The bottom layer consists of a grid of microresistors. On top of that sits a grid of sealed cells the size of poppy seeds, each holding a minuscule amount of fuel such as lead styphnate. These cells are covered with a thin membrane that can be broken by igniting the fuel.
To fire the cell, an electrical current is passed through the microresistor, which ignites the fuel, ejecting the propellant and delivering an impulse. Using a speck of the solid lead styphnate fuel, each cell generates about one newton of thrust (around a quarter of a pound) for about a millisecond. With arrays of such microthruster chips built into its walls, a microsatellite weighing no more than a kilogram could adjust its position in orbit with extreme precision.
The attraction of micropropulsion arrays is that they are simple to make and hard to break. Unlike most commercial rocket engines which use liquid fuels, these arrays use a "dry" technology. They have no moving parts, fuel lines, valves, sealants or the messy hydraulics of conventional rockets. Dr Lewis reckons that 1m microthruster cells could be manufactured on a single wafer of silicon.
The self-powered processor
The way MEMS devices are powered at present is far from satisfactory. Often the equipment for generating the necessary voltage is too large to make the MEMS device worth building. If micromachinery is to play the role enthusiasts predict, frugal chip-sized power sources are going to be needed in large quantities.
How about building a nuclear reactor on a chip? This has been suggested by two researchers at the University of Wisconsin. But the nuclear reactions that James Blanchard and Amit Lal propose using would be a lot safer than those used to drive nuclear power stations. A nuclear MEMS device would collect particles produced by radioactive decay. For example, a radioactive isotope such as nickel-63 releases so-called beta particles—a highly energetic form of electrons. Each electron carries about one nanowatt (a billionth of a watt) of energy. Harness a few thousand of these minute bursts of energy, and some useful work could be done.
In their experiments, the two Wisconsin researchers use a solution of the nickel isotope to fill sets of microscopic channels in a silicon plate. While the radioactivity emitted by the solution is not enough to damage the silicon, it is enough to produce a few microwatts of power. Though the output is minute, such a power source could carry on generating electricity unassisted for a long time. The half-life of nickel-63 is 102 years. So a nuclear-powered microgenerator would still have half a tank of fuel left after a century.
Being small and light enough, nuclear MEMS generators could be used to power microscopic aircraft. Kris Pister of the University of California in Berkeley wants to use nuclear MEMS devices to propel microscopic airborne surveillance devices around the sky. A floating network of silicon sensors the size of dust particles would monitor the temperature, pressure, moisture of the surrounding air, or even movements on the ground.
Science fiction? Far from. The reality is that microengines are fast becoming the power source of choice for the latest crop of MEMS devices—including the lab-on-a-chip being developed for the food and drug industries. The traditional battery has improved in leaps and bounds since laptops, cell phones and PDA s came on the scene. But it is still an order of magnitude or more too bulky for the new generation of micromachines. The fuel-cell on a chip is coming along fast and could well replace the rechargeable battery within a decade. But the theoretical limits of chemical energy mean that they will never be able to produce as energetic a punch as a MEMS generator. There is no escaping the fact that, ultimately, micromachines will have to have microengines to drive them.
Readers wishing to discuss issues raised in this article are encouraged to post their comments on the online forum at
Mar 14th 2002