A few years ago, just to show that they could do it, Paul Wright and his mechanical engineering students mounted a temperature sensor under a staircase at the University of California, Berkeley and fed its readings into the stairwell’s thermostat. It was not an especially difficult exercise—except that this sensor, which was about the size of a quarter, had no power cord or batteries. Instead, the device extracted the energy it needed from the vibrations that shook the wooden staircase as students clomped up and down between classes.
In Wright’s view, this kind of energy scavenging—with sensors and other electronic devices living off the land, so to speak—could open a new realm for technology.
“The 1990s marked this very interesting period in which devices for computing, communication, and sensing all became much cheaper and much, much smaller,” explains Wright. He’s an engineer who has worked in robotics and computer science and is currently chief scientist at the Center for Information Technology Research in the Interests of Society, a multicampus program supported by the state of California.
The most obvious result of the miniaturization was a wild proliferation of cell phones, personal digital assistants, MP3 players, and other portable gadgets. But in parallel, Wright says, “researchers were led to this picture of wireless sensor networks everywhere”—in effect, an electronic nervous system that reports on both the built environment and the natural landscape.
In the not-so-distant future, for example, bridges could tell us whether they had been damaged in an earthquake. Helicopter rotors and other high-stress machine parts could warn about developing metal fatigue. Office buildings could track the locations of their occupants, automatically adjusting the lights and air conditioning for maximum comfort and minimum energy use. Automobiles could talk to each other—and to the road—in an effort to avoid both accidents and traffic jams. Implantable sensors could continuously monitor blood-glucose levels and a host of other medical conditions. And webs of environmental sensors could monitor the health of remote ecosystems, tracking moisture, temperature, micronutrients, pollutants, and many other variables. All these developments would rely on networks of minuscule sensors (SN: 5/5/07, p. 282).
These applications and more are under active investigation, says Wright. Indeed, a wide variety of wireless sensors is already available commercially. But one of the biggest challenges continues to be power. Most of the potential applications call for so many sensors scattered so widely through a target area that it would be impractical to wire up each sensor individually—and ludicrous to run around changing batteries.
Thus the need for devices that can draw energy from their surroundings, says Wright. “The 1990s brought us computing, communication, and sensing. And now, I want to add the fourth thing—energy-scavenging devices—and make them as cheap as dirt.”
Energy scavenging is not a new idea. Self-winding wristwatches, in which a tiny mechanical oscillator extracts energy from the wearer’s arm movements, first appeared in the 1920s. And, of course, windmills and water wheels have been harvesting natural energy for thousands of years. But the current wave of interest in energy scavenging for microelectronics began in the late 1990s—initially because researchers were looking for a better way to power the newly devised portable devices.
In 1998, for example, Joseph Paradiso and his team at the Massachusetts Institute of Technology’s Media Lab demonstrated an energy scavenger embedded in the sole of a running shoe. It relied on the piezoelectric effect, in which crystals of certain materials produce a voltage in response to stress—in this case, from the impact of the wearer’s heel on the ground. And it worked, sort of.
A typical adult expends several hundred watts of power while walking, and 1,000 watts or more during strenuous exercise. But our bodies are also remarkably efficient. The MIT team found that a shoe that taps more than a tiny fraction of that energy flow gives the wearer the sensation of walking through mud. In the end, the Media Lab shoe generated only about 60 milliwatts—not enough to power an iPod, much less recharge a cell phone.
As a result of this experiment and others like it, energy-scavenging researchers soon shifted their focus from relatively power-hungry portable electronic devices to a new generation of far-more-thrifty gadgets made with microelectromechanical-systems (MEMS) technology.
The basic idea of MEMS, which dates to the 1970s, is to carve microscopic mechanical structures into the surface of a silicon wafer by using the techniques devised to create microprocessors. By the 1990s, MEMS researchers had produced all manner of gears, springs, cantilevers, channels, and the like, and a first generation of commercial MEMS devices was reaching the market. Applications included the microscopic nozzles of inkjet printers, the chip-size accelerometers that trigger the deployment of an automobile’s airbags in a collision, the micromirrors of a digital light-processing display, and the microchannels that move fluid around for analysis in an integrated lab on a chip.
Moreover, the silicon wafers of MEMS devices could easily be integrated with microelectronic circuitry. If sensors, manipulators, and computational smarts—as well as microscopic energy-scavenging devices—could be etched into the same chip, they’d add up to microscopic robots, sometimes referred to as motes or smart dust, that could form the working elements of self-powered, wireless sensor networks.
From an energy-scavenging standpoint, however, the great advantage of MEMS sensors is that they typically require only about 100 microwatts of power—a thousandth of what portable consumer electronic devices typically need. Such minuscule quantities of energy abound in the environment: in vibrations, temperature gradients, sunlight, and so on.
The challenge is to make effective use of that energy.
The first thing to keep in mind is that there is no all-purpose solution, says Steve Arms, founder and president of MicroStrain, a manufacturer of MEMS-based sensors in Williston, Vt. “The biggest challenge is really understanding the environment you’re working in and seeing what energy is available.”
Take solar energy, for example. It’s notoriously variable and unpredictable, thanks to clouds and the day-night cycle, and obviously unsuitable for sensors mounted indoors or implanted in a person’s body. But a tiny photovoltaic cell might be just the thing for an environmental sensor that is regularly exposed to bright sunshine.
Now, consider thermal energy. The only way to extract it is when heat flows from hot material to cold. Ambient temperatures rarely vary enough from point to point to make that a winning proposition. But again, thermal-energy extraction might be valuable for certain applications.
Wireless Industrial Technologies, an Oakland, Calif., company cofounded by Wright, uses energy scavengers that take advantage of the Seebeck effect, in which certain metals develop voltages when one end is hotter than the other. “We’ve mounted the devices in an aluminum-smelting plant, between the outside of the smelter and a cooling fin. We can get a 50°C difference,” Wright says. That’s enough to power sensors that monitor many of the plant’s operating parameters.
Then there are vibrations, which are quite variable—but ubiquitous. “You do have to be a little bit clever about where you place the sensors,” says Wright. “But, for example, if you go through a commercial building, you find that the big windows vibrate a lot because of buses and trucks going by. You also get lots of vibration on air-conditioning ducts, on raised floors, and around heavy doors.”
In certain settings, notes Arms, the vibrations can be strong indeed. He and his colleagues are currently working with Bell Helicopter of Fort Worth, Texas on a project for the U.S. Navy.
To avoid problems due to metal fatigue, the moving parts on a helicopter, especially the blades, are currently replaced on a schedule determined by how many flight hours the vehicle has logged. “But it would be much cheaper and safer to track the actual fatigue and replace the blades only when you need to, which might be longer than the schedule called for—or sooner,” says Arms. “So, this past February, we did a successful flight test of a wireless strain sensor that got all its power from vibration during the helicopter’s normal operation.”
Such structural health monitoring is critical for all kinds of industrial machinery, adds Shashank Priya of the University of Texas at Arlington and host of an annual series of workshops on piezoelectric-energy scavenging. Indeed, structural monitoring could be the most important commercial application of vibration-powered sensors.
“There are 3,000 to 6,000 sensors inside a modern jet fighter or commercial aircraft,” says Priya. “Currently, all of them are wired in for data, often with batteries for power. But that’s tough: The aircraft is full of wires, the wires and sensors have to be manually checked very frequently, and it’s very tedious to repair them if something goes wrong.” If the sensors were wireless and could power themselves, many problems would disappear.
The efficiency challenge
Standing in the way of such applications, unfortunately, is a second big piece of the energy-scavenging challenge: capturing tiny bits of ambient energy efficiently.
“Most of the devices right now are only 10 to 15 percent efficient,” says Wright. That makes it tough for them to compete in the marketplace against, say, long-lived batteries.
A typical piezoelectric scavenger built using MEMS techniques has an array of microscopic cantilever beams—in effect, tiny diving boards that twang whenever a vibration comes through. Those oscillations, in turn, stress the piezoelectric crystal and produce the voltage. But when the cantilevers are this small, their oscillations occur at many thousands of cycles per second. Because ambient vibrations typically have frequencies of a few hundred cycles per second, they don’t efficiently set the cantilevers into motion.
Much research has gone into improving that fit, largely by refining the device’s structure to shift the cantilevers’ natural resonance down to more-useful frequencies, says Wright.
Meanwhile, researchers have also been looking at more-efficient—and less-toxic—alternatives to the most commonly used piezoelectric material, called both lead zirconium titanate and PZT. Its lead content makes its use especially problematic in implantable sensors.
At the Georgia Institute of Technology in Atlanta, for example, Zhong Lin Wang and his group are developing a novel nanoscale energy scavenger based on zinc oxide—a material best known for its use in sunscreen. On the face of it, zinc oxide is not nearly as good a piezoelectric as PZT. “But it is both a piezoelectric and a semiconductor, which is very rare,” says Wang.
To demonstrate how these properties can work in concert, he and his team grew a forest of upright zinc oxide nanowires, each of them a perfect crystal. The researchers then lowered an array of sharp nanoscale electrodes fabricated by similar techniques, leaving just enough space so that when the nanowires flex, they touch the electrode tips.
When ambient vibration or other mechanical energy deflects the nanowires, they develop piezoelectric voltages that move charges within the semiconducting material. During the periodic contacts between the nanowires and the electrodes, those charges move into the electrodes.
This effect could be the basis of energy scavengers with efficiencies as high as 30 percent, Wang predicts. “I have a high hope that we will be able to market commercial zinc oxide nanogenerators within 3 years,” he says.
Onboard energy management
Efficiency is also a major goal of electrical engineers and software designers working on a third piece of the energy-scavenging challenge—managing how the energy is employed by the sensor package itself. One frequently used strategy is to have the devices spend most of their existence in sleep mode, where they can survive on just the barest trickle of power. They have to wake up for only a tiny fraction of a second every now and then to take a quick instrument reading and, if necessary, beam back a few bits of data.
Nonetheless, notes Priya, that “beaming back” part continues to be tough. “Present-generation sensors are very efficient and consume only 50 to 100 microwatts. But a transmitter consumes on the order of 50 milliwatts,” he says.
To give the transmitter enough power for its occasional bursts of activity, the device would need to accumulate scavenged energy in some sort of long-lived battery. The relatively new technology of thin-film lithium-ion batteries is especially appealing for such applications, says Arms. “They are paper-thin and flexible, and they can go through an essentially infinite number of recharging cycles,” he says.
The final step in the energy-scavenging challenge is to integrate all the pieces into robust, complete systems (SN: 5/5/07, p. 282).
“The long-term dream is that everything will be fabricated on a single wafer,” says Wright. Then, the devices could be produced en masse. Wright notes that, for example, Elizabeth Reilly of the Berkeley group is already making such integrated devices in the lab with MEMS processes.
The bottom line, Wright adds, is that energy-scavenging technology has a long way to go—but it is moving fast.