Roughly two-thirds of the money spent on power might as well be burned, as about two-thirds of all energy used by devices such as lights and computers is lost as heat. But prospects for reclaiming some of that wasted energy have improved, now that scientists have devised a so-called thermoelectric material that crosses a key practical threshold for generating electricity from heat.
Much as the photovoltaic compounds in solar cells generate electricity from light, thermoelectric materials convert heat to electricity. That raises the possibility of, say, using some of the waste heat from the gasoline engine in your hybrid car to charge the car's battery. "The dream of the field is to harvest waste heat from as many situations as possible—vehicles, factories, tankers," says Mercouri Kanatzidis, an inorganic chemist at Northwestern University in Evanston, Illinois. "You'd then put that energy back into where you harvested it from—for instance, into increasing the mileage on your vehicle." There is, of course, a limit, as the second law of thermodynamics says that it's impossible for any device to run without generating some waste heat.
Within a solid, heat can be thought of as tiny quantum mechanical vibrations of atoms and molecules, known as "phonons." Thanks to their atomic-scale structure and electrical properties, thermoelectric materials convert this vibrational energy into the flow of electrons. If one end of a piece of thermoelectric material is warmer than the other, a voltage difference will emerge between the two ends, allowing the material to generate a current, a bit like a battery.
Scientists rate the performance of a thermoelectric material by a measure known as ZT, which accounts not only for the stuff's ability to produce a voltage, but also its ability to conduct electricity (which should be high) and its ability conduct heat (which should be low). The best ZT values researchers had achieved were between 1.6 and 1.8, but researchers hoped to reach a value of 2, at which point applications of thermoelectrics would become more practical. "A ZT in the range of 2 and above represents an overall heat-to-electricity conversion efficiency in the 12 to 17% range, similar to what you'd see with industrial photovoltaics," Kanatzidis says. (A ZT of 2 is almost twice as efficient as a ZT of 1.)
The new thermoelectric material consists primarily of lead and tellurium; past studies found that lead telluride was the best thermoelectric system at the kind of high temperatures one might find in engines and other hot spots. The researchers adopted three different techniques for soaking up energy from phonons. Within the material, grains of semiconducting lead telluride that are hundreds to thousands of nanometers wide absorb phonons of longer wavelengths. Also, precipitates of strontium telluride 2 to 10 nanometers wide target shorter wavelengths. Finally, trace amounts of sodium injected within the material's crystalline structure go after the shortest wavelengths. As a result, the material achieves a world-record  ZT of 2.2. "That's conservatively between 15 and 30% more efficient than the previous record-holder," Kanatzidis says. The scientists detailed their findings online today in Nature.
"Creating new materials with enhanced thermoelectric properties is always a challenge, since it requires properties nature doesn't always want to give you in a single material—high electrical conductivity and low thermal conductivity, for example," said Donald Morelli, a physicist at Michigan State University in East Lansing who was not involved in this study. "The approach they use here, based on engineering a material's structure on multiple length scales, is one that's really generalizable—there's no reason why you couldn't apply it to any semiconductor material"—the sort of stuff used in microchips.
Kanatzidis and his colleagues now want to design materials with even higher ZT values, such as 2.5 or 3. (A ZT of 3 is about 2.4 times as efficient as a ZT of 1.) They also want to develop thermoelectric materials that don't require tellurium, which is about as rare as platinum. Cheaper candidates include lead selenide and lead sulfide.