Chang made carbon nanotubes and boron-nitride nanotubes between 10 and 40 nanometers wide. He placed individual nanotubes in a test chamber, with each end of the tube bonded to silicon and platinum-based electrodes that act as either heaters or sensors. Then he deposited a platinum compound unevenly along the length of the nanotube so that the tube had more mass at one end than at the other. Finally, he sent a known amount of power through the heater end and measured the change in temperature at both electrodes to see how much heat was passing through the nanotube. As it turned out, as many as 7 percent more phonons were traveling from the high-mass end to the low-mass end than in the opposite direction.

This small efficiency is not enough for practical applications. But what's more important at this stage is the solid proof that the thermal-rectifying effect exists, says Giulio Casati, a professor of physics at the University of Insubria, at Como, Italy, who, along with Peyrar, originally proposed the idea of a thermal rectifier. "It's the first step," he says. "When [engineers] built the first electrical diode, the efficiency was very low. So it'll take time."

Majumdar says the next step is to explore various nanotube geometries as well as the platinum compound loaded on the nanotubes. "Could we change that material or could we change the geometry and thereby increase rectification?" he asks. "That's still up for grabs right now."