That may sound like a fairly simple engineering challenge, but it isn't, despite the well-known fact that heat can be used directly to produce an electric current. If you heat one end of a wire, for example, electrons will flow toward the cold end of the wire, and that's electricity. But for that process to continue, one end of the wire must remain hot, and the other cold. And that's a problem because most materials that conduct electricity also conduct heat.
That's especially true for silicon, so most scientists had concluded that it was not a very good candidate for thermoelectric conversion. And that's not what Majumdar and Yang and their colleagues had in mind when they immersed a silicon wafer in a chemical solution. They were looking for an easier, and cheaper, way to grow tiny wires, called nanowires, on a silicon wafer. These microscopic wires have high potential in a wide range of uses, but it's not easy to produce them.
The researchers thought they had hit pay dirt when a microscopic forest of nanowires grew on the surface of their submerged wafer. But when they looked closer, they found that the wires were substandard. Instead of the smooth surface they had expected, the surface of their nanowires was rough.
But thinking the flawed wires might have some purpose, they tried to use them in a photovoltaic device.
"It didn't work," Majumdar said.
Intrigued, they tried to use their wafer as a thermoelectric device. And that's when they found success. The rough nanowires did allow current to flow from a heat source toward a cold source, but astonishingly, the heat did not also flow from hot to cold. That seems contrary to all that is known about silicon, and it was a bit baffling to the researchers.
Although more work needs to be done before the physics is understood, Majumdar has a hunch about what's going on.
Heat travels through a material like sound waves travel through air. He compares the nanowires to a corrugated tube. If you were to shout into the tube, some of the sound waves would bounce off the corrugated ridges and valleys, and eventually nearly all of the sound would be lost. Similarly, the heat waves may simply be bouncing off the rough contours on the surface of the nanowires, allowing electricity to flow, but dissipating the heat.
So the roughness of the nanowires, which was not planned, turns out "to be the key," Majumdar said.
This "breakthrough," as Majumdar describes it, won't solve the world's energy problems, and it won't end global warming, but it could help. And since there is already an infrastructure in place, the $100 billion semiconductor industry, it may come sooner rather than later. All because of a timely bit of serendipity.
Lee Dye is a former science writer for the Los Angeles Times. He now lives in Juneau, Alaska.