NASA's latest rover on Mars depends on a sandwich of semiconducting material that can turn heat into electricity. In the case of Curiosity, the steady radioactive decay of plutonium 238 warms such thermoelectric material and turns roughly 4 percent of that heat into a steady flow of electrons. A similar radioisotope thermoelectric generator (RTG) on the moon's Sea of Tranquility is still working after decades, as are the RTGs in the two Voyager spacecraft launched 35 years ago; such enduring reliability is the main reason NASA employed the inefficient technology. Now researchers have discovered a way to at least double the efficiency of such power generators—suggesting that thermoelectrics might find a home in applications outside of aerospace and back here on Earth.

The most common core of new and old thermoelectrics is a compound called lead telluride. When exposed to heat on only one side—whether it be from a radioactive isotope or another source—it will induce an electric current as long as the temperature differential is maintained. The challenge of improving thermoelectrics has been to keep heat from transferring across the material without also interfering with its ability to conduct electricity.

Chemist Mercouri Kanatzidis of Northwestern University and his colleagues report in Nature on September 20 that by precisely engineering the material from the atomic to the individual grain scale, the thermal conductivity of lead telluride can be impeded without affecting its electrical conductivity. The result is a material that can convert at least 8 percent of the heat into electricity—and could theoretically convert as much as 20 percent. (Scientific American is part of Nature Publishing Group.)

The researchers first melted the lead telluride and then froze it, creating nanoscale crystalline structures out of the atoms. These precisely oriented nanostructures scatter the medium wavelength vibrations, or phonons, that carry heat while allowing electrons to pass unobstructed.

But longer wavelength phonons continue to pass through as well, because their wavelengths are longer than the size of the nanostructures. So Kanatzidis and his colleagues went further, grinding the nanostructured material into powder. The powder was then subjected to spark plasma sintering—squeezing the powder while also passing "a very large amount of [electrical] current," in the words of Kanatzidis, through it briefly—to consolidate the grains into a larger block. Because the sintering occurs so quickly, the material does not melt but does compact, making it hard enough to be cut and manufactured into the core of a thermoelectric device. And these grains then effectively block the longer wavelength heat while still allowing electricity to flow.

This combination of what Kanatzidis calls "panoscopic" processes results in a lead telluride material that is more than twice as efficient at converting heat into electricity at high temperatures. "It's pretty significant and it makes the whole thing smaller," Kanatzidis notes.

Or, as chemist Tom Nilges of the Technical University of Munich wrote in the same issue of Nature on the new manufacturing process he likens to a matryoshka doll, consisting of smaller and smaller units of material nested within one another, "this combined approach improves the thermoelectric performance of lead telluride to previously unattainable levels."

At those levels such thermoelectric devices might become practical in harvesting some of the exhaust heat from vehicles—such as marine tankers or trucks—and turning it into electricity. BMW and Ford are already testing similar thermoelectric material in cars. Or the devices could be used in high-heat metallurgical or glassmaking industries. And scientists at the Massachusetts Institute of Technology have even used such thermoelectric materials to build a device to turn the sun's heat more directly into electricity, rather than employing the vast arrays of mirrors of a conventional solar-thermal power plant.

Of course, both lead and tellurium are toxic, but nontoxic alternatives, such as zinc oxide, might prove feasible in future. At the very least, the next robotic rover to land on a foreign world could have a lot more juice.