Fossil fuels power modern society by generating heat, but much of that heat is wasted. Researchers have tried to reclaim some of it with semiconductor devices called thermoelectrics, which convert the heat into power. But they remain too inefficient and expensive to be useful beyond a handful of niche applications.

Now, scientists in Illinois report that they have used a cheap, well-known material to create the most heat-hungry thermoelectric so far. In the process, the researchers say, they learned valuable lessons that could push the materials to the efficiencies needed for widespread applications. If that happens, thermoelectrics could one day power cars and scavenge energy from myriad engines, boilers, and electrical plants.

Thermoelectrics are slabs of semiconductor with a strange and useful property: heating them on one side generates an electric voltage that can be used to drive a current and power devices. To obtain that voltage, thermoelectrics must be good electrical conductors but poor conductors of heat, which saps the effect. Unfortunately, because a material's electrical and heat conductivity tend to go hand in hand, it has proven difficult to create materials that have high thermoelectric efficiency—a property scientists represent with the symbol ZT.

Several years ago, researchers led by Mercouri Kanatzidis, a chemist at Northwestern University in Evanston, Illinois, discovered an impressive new thermoelectric material called lead telluride (PbTe), which had a ZT value of 2.2. That was reasonably close to the ZT of 3 that most researchers consider the minimum for widespread applications.

Intrigued, Kanatzidis and his colleagues started testing PbTe’s chemical cousins. One was a shiny silver material called tin selenide (SnSe). Decades earlier, researchers had found that it was too poor an electrical conductor to be worth trying as a thermoelectric. But tin and lead belong to the same group in the periodic table, and tellurium and selenium both are members of another group. “It was a curiosity that we wanted to explore,” Kanatzidis says.

So Kanatzidis and colleagues at Northwestern and the University of Michigan, Ann Arbor, decided to take another look at tin selenide. The researchers synthesized a bullet-sized sample of SnSe and cleaved pieces of it along three different orientations of the crystal’s atomic planes, known as the a-, b-, and c-axes—a standard technique for analyzing the properties of materials. They then charted the thermal and electrical conductivity of each sample across a wide temperature range. The b-axis sample turned out to have a better-than-expected electrical conductivity and a very low thermal conductivity to boot. Those properties gave the material a ZT of 2.6, the best value ever measured. The key to the ultralow thermal conductivity, Kanatzidis says, appears to be the pleated arrangement of tin and selenium atoms in the material, which looks like an accordion. The pattern seems to help the atoms flex when hit by heat-transmitting vibrations called phonons, thus dampening SbSe’s ability to conduct heat. The researchers report the results today in Nature.

“I’m amazed,” says Joseph Heremans, a physicist at Ohio State University, Columbus, who wasn’t connected to the research. “This is a fantastic result for the field.” In addition to marking a big step toward thermoelectrics with a ZT of 3, Heremans says, the new material offers lessons on how to get there. Most likely, he says, researchers will try to boost the semiconductor’s electrical conductivity by spiking it with trace amounts of “dopant” atoms, while preserving the key accordionlike atomic arrangement. If anyone succeeds in producing a high-ZT material, Heremans says, it could lead to new, cheaper hybrid car engines in which the internal combustion engine doesn’t power the car, but rather generates heat that thermoelectric devices convert into electricity to power an electric motor. For now, that’s still a vision of the future. But it’s one that now appears closer than ever before.