Last December, a team of managers, scientists, and technicians from Texas Instruments (TI) trekked from their corporate research labs in Dallas to a meeting at George Lucas’ sprawling ranch in the hills west of San Rafael, Calif. Lucas was present. So was Rick McCallum, producer of “Star Wars: Episode I-The Phantom Menace,” then still half a year from opening day. The movie moguls had summoned the TI staffers to audition a digital projector, built by the company, that could change how movies are shown, replacing Hollywood’s beloved canisters of film with semiconductor chips.
Lucas was impressed. He called back the TI scientists. “He asked if we were interested in unveiling the digital cinema projection system at a second opening of ‘Phantom Menace,’” recalls Larry Hornbeck, a soft-spoken TI physicist who has been working for the past 20 years on the digital micromirror device (DMD) that is at the heart of the new projector.
The rest is cinematic history. In mid-June, a month after the opening of “Phantom Menace” drew millions of fans, the much-anticipated movie debuted a second time at a theater outside Los Angeles and one just across the Hudson from Manhattan. Unlike the initial opening in May, however, not a frame of film snaked through the projectors. Instead, a 360-gigabyte digital file stored on 20 hard disks fed the special TI projectors lighting up the big screen.
Making it all possible was a microelectromechanical system (MEMS) featuring millions of mirrors-each roughly the size of one cell in the human body. Transistors adjust each mirror individually to one of two possible angles. Set one way, light from a projection bulb reflects from the micromirror through a set of focusing lenses and onward to the screen; tilted the other way, the light dumps into an absorber inside the projector. It’s akin to a sports arena filled with millions of fans holding cards they can flip in perfect unison to construct an enormous image. Only with the TI projector, the high-resolution pictures can change every 5 microseconds (far faster than the eye can discern). By using sophisticated software and alternating beams of red, green and blue light, the system can produce more than 1 billion colors-plenty even for Lucas’ cinematic fantasies.
A Star Is Born
The special screening of “Star Wars” may or may not have served as a preview to Hollywood’s filmless future (Lucas also chose a competing digital projection technology developed by CineComm Digital Cinema to show the movie at a separate pair of theaters). But it’s the most glamorous sign to date that MEMS devices, often too small to be seen by the human eye, are poised to change everything from how we watch movies to the size of the cell phones we carry.
The miniature silicon machines move and work, unlike semiconductor chips used in microelectronics, in ways remarkably reminscent of macroscopic objects. And, after decades of development and hundreds of millions of dollars in investments, the micro-sequel to the original machine revolution is gaining momentum. Dot-sized motion sensors are now used in most automotive air bags, and ultrathin silicon membranes are used to measure blood pressure inside human hearts; coming attractions will feature tiny mirrors that can switch light between optical fibers to greatly enhance the efficiency and reliability of optical communications, and resonators that could shrink wireless communication, making Dick Tracy’s wrist phones a reality.
“The technology has methodically advanced during the last 10 to 15 years, and it’s just hitting its commercial stride,” says MEMS researcher and entrepreneur Mehran Mehregany of Case Western Reserve University. Market analyses put MEMS annual sales at several billion dollars and predict the number will grow to between $5 billion and $10 billion within five years.
Those numbers have helped create a gold-rush mentality among researchers-and some investors. There are an estimated 10,000 scientists working on MEMS at 600 universities, government labs, big companies and tiny startups, according to Roger Grace, a San Francisco-based consultant. Fueling the fever are tales of MEMS startups striking it rich. Most notably, in 1997 data storage giant Seagate Technology bought San Jose, Calif.-based Quinta for $325 million. The startup, founded just two years earlier following a Silicon Valley cafe meeting, is developing a system using micromirrors-somewhat similar to Hornbeck’s-in a laser guidance system for next-generation disk drives.
Alongside the enthusiasm, MEMS has gained a reputation as a technology better suited to building tiny playthings than to constructing moneymaking machines. In the early 1980s, researchers first realized silicon chips could be as good for making very small mechanical parts as they were for making electronic circuits. Scientists began developing microfabrication techniques to make Lilliputian wheels and gear trains, motors that could push shafts, valves for controlling microflows of fluid, and mirrors that pop up into the path of a laser beam. But MEMS labs retained the aura of high-tech curio shops most memorable for showcasing the world’s smallest versions of everything from guitars to cars. “There was too much gimmicky press,” acknowledges Karen Markus, vice president of Cronos Integrated Microsystems in Research Triangle Park, N.C.
While microscopic pictures of mites and ants towering over clockwork-like structures are irresistible to journalists, some venture capitalists find them distinctly resistible. “My feeling is that most MEMS work is ill-conceived,” says Greg Blonder, a former MEMS researcher turned venture capitalist, or as he calls himself, “entrepreneur in residence,” at AT&T Ventures in Basking Ridge, N.J. Blonder contends that many examples of MEMS amount to little more than miniature Rube-Goldberg machines, and he says that MEMS projects are often answers searching for problems that have far simpler solutions. Says Blonder: These projects “add no real value but add complexity.”
But that doesn’t mean Blonder dismisses all MEMS projects. He simply argues that MEMS devices must be conceived of differently than macroscopic machinery. MEMS operate on a scale of micrometers and millimeters, and in that small world things can behave somewhat differently. There are surprising scaling effects-these effects, for example, allow ants to lift up to 50 times their own weight. “Where the real excitement is going to be is in MEMS that leverage the strengths of physics of those dimensions,” Blonder says. Sensors, relays and other micromechanical devices capable of using extremely slight pressure and temperature fluctuations to move parts are promising, he adds. So are devices in which small sizes can speed up chemical reactions and more efficiently dissipate heat.
Blonder’s cautious, discriminating approach has done little to dampen enthusiasm among researchers. The success that semiconductor chipmakers have had in shrinking microelectronics serves as an omnipresent reminder that smaller is better-and more lucrative. Microelectronics prove that you can change society on a large scale through miniaturization, mass production and cost reduction, argues Neal Barbour of Draper Laboratories in Cambridge, Mass. In his view, microprocessors and memory chips are mere hints of a much more generalized trend of miniaturizing technologies. “MEMS is fundamental, just like electronics,” says Al Pisano, director of the Defense Advanced Research Projects Agency’s MEMS funding program, which has an annual budget of $50 million. “MEMS will grow just like electronics. And it will become just as ubiquitous.”
Cheap as Chips
One reason for the excitement is that while microelectronics semiconductor chips are great at logic and memory, they are a brain without a body. “Computers think and think and think. But MEMS are becoming the eyes, ears, noses, mouths, hands and feet of computers,” says Markus. Adds Barbour: “All of the electronic components end up passive, but MEMS can respond to all kinds of inputs-chemical, light, heat, pressure, vibration, acceleration-all of the things that just about everybody needs to measure in just about every physical system that we have.”
These advantages of MEMS would be enough to entice researchers, but not enough to get MEMS devices onto the market. Yet another factor has also entered the picture: These systems are no longer exotic items that take microengineering specialists months to fabricate. TI’s micromirror arrays are, for example, made using lithographic techniques adapted from the microelectronics industry. Technicians start with silicon wafers, spin on thin coatings of polymer photoresists (a photosensitive material), expose plots of the photoresist to light through a stencil-like mask, and wash away the exposed photoresist to reveal a pattern of the underlying wafer that matches the pattern on the mask. After they expose the naked wafer surface, MEMS makers then etch into, around and underneath the surface, diffuse ions into the silicon, or deposit materials such as aluminum onto it. Clever sequences of masks, etching and deposition yield tiny 3-D structures that move on command.
Make no mistake: It’s still not easy. But the reliance on standard fabrication tools means manufacturing technology is already in place and, once you design a MEMS device, you can potentially turn them out as cheaply as semiconductor chips. At Analog Devices, headquartered in Norwood, Mass., some 1 million tiny accelerometers are fabricated every month, according to Jeffrey Swift, director of engineering for the company’s micromachined products. Buy a car today, and there’s almost a 50-50 chance that one of Analog Devices’ accelerometer-based sensors (each about the size of the period at the end of this sentence) will be inside the air bag systems.
Prior to micromachined accelerometers, motion sensors in air bags required up to five fist-sized components, each costing about $18. Analog Devices and several competitors sell MEMS accelerometers for less than $10 apiece, reflecting the companies’ ability to make them in huge batches. The accelerometers are relatively simple-suspended rectangular slabs of silicon with fingers extending out to form what looks like a double-sided comb. The fingers of these combs mesh with silicon fingers machined into the surrounding silicon framework. The normal motion of a car, as well as the violent and jerky motion during the split second of a crash, instantly sets the suspended accelerometer in motion. The overlapping areas of the meshed silicon fingers change, which causes instantaneous changes in the structure’s electrical capacitance. Those electrical changes then feed into circuitry programmed to discern potentially deadly crashes from potholes; when appropriate, the circuitry triggers the release of the air bag.
Analog Devices’ engineers are developing MEMS for other emerging automotive applications such as side-impact air bags. And Swift, a father of three boys, including a new driver, suggests another possibility: “What if you had a sensor in the car that your teenager was driving that would tell you how many g forces the car experienced? Did he peel out or do a harsh stop or go around corners at really high speeds?”
In fact the success of MEMS may depend more on engineers being able to think up new uses for them than it does on the skill of microfabrication labs. Kurt Petersen, widely considered one of the fathers of MEMS because of a seminal paper he wrote in 1980 at IBM Research in San Jose, Calif., is, for one, focusing his energy on Cepheid, which he co-founded three years ago in Sunnyvale, Calif. “Our role in life is DNA analysis,” says Petersen. And he’s convinced MEMS are going to prove useful.
One of the most time-consuming and least automated tasks of DNA analysis is the preparation of specimens-extracting, concentrating and purifying the genetic material so that it can be copied and sequenced. “People have not been able to extract and purify DNA on a microscale,” points out Petersen. Enter MEMS. Using a technique called deep reactive ion etching, in which ions chew straight down into exposed areas of a silicon wafer (as opposed to the surface micromachining used to make Texas Instruments’ tiny micromirrors and Analog Devices’ accelerometers), Petersen and his colleagues have created microfluidic chambers to capture and concentrate DNA and RNA. One design consists of a square array of 1,600 pillars, each 10 micrometers in diameter and 10 micrometers from its neighbors; as biological samples flow through this forest of silicon, DNA (or RNA) sticks to the pillars. The captured nucleic acid can later be released by an appropriate buffer; once separated, DNA travels through an exit port to downstream modules for analysis.
Cepheid expects to market a handheld instrument that can accept virtually any biological sample, extract its DNA, and then amplify and detect nucleic acids previously identified to be of interest. Cancer screening, as well as detection of pathogens and biological warfare agents, is among the applications. Full commercialization is expected in several years.
Cancer diagnosis is serious stuff and enormously important. But other uses for MEMS are lighter at heart. At the University of Michigan’s MEMS research center, Clark T. Nguyen envisions a Dick Tracy future based on MEMS resonators and frequency filters for cell phones and other communications gadgets. These are the components that ensure transmission takes place at specific frequencies and allow receivers to tune into particular frequencies plucked from the cacophony filling the airwaves. Unlike the pinkie-nail-sized quartz crystal resonators used in cell phones, whose oscillations are in the form of mechanical waves traveling back and forth along the solid crystal’s atomic framework, MEMS resonators have moving parts more akin to pendulums. Yet thousands could fit into the same space taken up by one of their crystal rivals.
As researchers like Nguyen miniaturize radio frequency components for cell phones, pagers, Global Positioning System receivers and other technologies, devices can become more selective in the frequencies they receive or transmit. Such selectivity means they require less power. “We are looking at a process that can take the whole cell phone and fit it onto a wristwatch, or a ring on your finger,” Nguyen says.
Others predict they are on the verge of using MEMS to transform the communication infrastructure based on optical fibers and other light-based technologies. Considering that the light-carrying core of optical fibers is about 9 micrometers in diameter, it’s no surprise that systems engineers at places like Lucent Technologies would look to diminutive devices to help them steer light through fiber networks. “We believe MEMS is going to really revolutionize how photonic switching gets done,” says David Bishop, head of microstructure physics research at Lucent’s Bell Laboratories in Murray Hill, N.J. Some of the first of these microswitches for light could show up in the communication network in the next year or so, says Bishop.
“For the moment, optical switches are heavy and expensive pieces of equipment,” explains Bishop. “Some involve just taking the end of a fiber, connecting it to a motor and you move it [from one fiber to another] that way.” Instead of moving the fibers, MEMS promises more elegant, cheap and reliable ways of optical switching using tiny mirrors and lenses-in other words, by guiding the light. “The nice thing about photons is they are small and don’t weigh much, so you can use a micromachine to move them around,” says Bishop.
In one MEMS design, Bishop and colleagues have fabricated a “thermally deformable micromirror” that can change its focal distance. Smaller than a poppy seed, it looks like a radar dish with eight individually tiltable triangular wedges (each made of gold-coated silicon). With this design, light coming from eight fibers can be precisely recombined (multiplexed) into downstream fibers positioned at several of the adjustable mirrors’ focal points.
Devices like these could help reduce the vulnerability of photonic networks to failure. If an optical fiber gets severed in the middle of the night, explains Bishop, you would like to be able to reconfigure the path of light in that network without having to send a truck out to fix the fiber. As Bishop sees it, pressing a button that sends a little jolt of light or electricity to the invisible wires of a poppy-seed-sized mirror might be all it will take to reroute the light and keep the lines open.
If the delicate manipulation of DNA and photons seems a likely match for tiny machines, consider the research going on at the gas turbine laboratory at MIT. This is home to engineers expert on the multi-ton turbines and jet engines that have powered industry for decades. But these days it is also home to one of the most promising new twists in tiny machines-power MEMS. What if you could use the energy from tiny combustion chambers to drive microturbines and microgenerators? Or if you could build rocket engines the size of a coat button?
A group of several dozen MIT scientists are trying to do just that. One project involves building a microjet that runs on hydrogen and that could be used to power a 15-centimeter-long airplane. “The whole idea was somewhere between silly and fantastic when we started about 10 years ago,” says Alan Epstein, director of the Institute’s gas turbine lab and one of the originators of the power MEMS effort. “But we ran the numbers. They said this could work.” Since then, Epstein and his collaborators at MIT and elsewhere have chipped away at many of the necessary parts-little fuel injectors, combustion chambers and silicon microturbines that look something like high-tech pinwheels. Says Epstein: “This is not vaporware.”
The microjets could end up becoming part of a fleet of surveillance air vehicles several inches long that military planners would like to see in their bag of intelligence tricks. But the implications of power MEMS go far beyond that-it could solve a fundamental challenge facing engineers as they try to build smaller and smaller devices (see table “MEMS Mania”). How do you power them? Conventional batteries have not kept pace with the miniaturization trend. “The most important economic possibility is to use these little engines like batteries,” suggests Epstein.
Just as power companies use massive gas turbines to drive electrical generators, Epstein says microturbine gas engines could become an alternative to batteries for supplying portable power. In a package of a few cubic centimeters, he says, a microturbine fueled by a tiny aluminum tank of hydrocarbon fuel could supply 20 or 30 times the energy of a conventional battery. Laptop computers could start feeling more like pads of paper and cell phones could last weeks without needing to be rejuiced. Such a MEMS power source is still at least several years off. But, says Epstein, “we’re convinced it can work.”
Battery manufacturers like Duracell may not need to look over their shoulders-at least not yet. But the progress of Epstein’s microturbines from theoretical musings a decade ago to prototypes today shows that once-arcane MEMS projects could end up transforming even a multibillion-dollar business like batteries. And the ambition of Epstein and his colleagues at MIT to use tiny spinning turbines to power the next machine revolution shows just how strong an impact MEMS has already made on the engineering mindscape.
How fast the revolution of tiny silicon machines will take to change the technology that we all use and rely on is still anyone’s guess. But if George Lucas could make a squat little robot into a worldwide movie star, he-and thousands of scientists-just may be able to help do the same for microscopic machines. Only this time, it’s not science fiction.
Company Goal Status Analog Devices Micromachined accelerometers and actuators Commercial products BF Goodrich Aerospace sensors Research Caliper Technologies (1995 startup: Mountain View, CA) Lab-on-a-chip for microscopic fluid handling Commercial products Cepheid (1996 startup; Sunnyvale, CA) Micro-chips for processing of DNA samples Commercial products Goodyear Tire Intelligent tires that sense pressure Prototypes Hewlett-Packard High-capacity data storage Research Honeywell Polymer actuators capable of affecting macroscale objects Research IBM Terabit data storage devices Research Lucent Technologies Components for optical communications Prototypes Rockwell Radio frequency switches for communication systems Research Seagate Technology Optical-based data storage devices Prototypes Texas Instruments Digital projection systems Commercial TRW Digital propulsion for micro-satellites Research Xerox PARC Microscopic controls for macroscopic systems Research Silicon Light Machine (1994 startup; Sunnyvale, CA) High-resolution displays Product development Leading MEMS projects by industrial groups in the United States. Source: Company reports and Defense Advanced Research Projects Agency (DARPA).
The new version of GPT-3 is much better behaved (and should be less toxic)
OpenAI has trained its flagship language model to follow instructions, making it spit out less unwanted text—but there's still a way to go.
A horrifying new AI app swaps women into porn videos with a click
Deepfake researchers have long feared the day this would arrive.
We can’t afford to stop solar geoengineering research
It is the wrong time to take this strategy for combating climate change off the table.
Meet Altos Labs, Silicon Valley’s latest wild bet on living forever
Funders of a deep-pocketed new "rejuvenation" startup are said to include Jeff Bezos and Yuri Milner.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.