Others are making solar cells with different nanostructures. Harvard University chemistry professor Charles Lieber has made nanowires consisting of a silicon core and different concentric silicon layers. Peidong Yang, a chemistry professor at UC Berkeley, has made dye-sensitized solar cells with zinc oxide nanowires. These nanowire solar cells have reached efficiencies of 4 percent.

Javey and his colleagues make the nanopillar cell by first anodizing aluminum foil. This creates a periodic arrangement of 200-nanometer-wide pores, which act as templates for cadmium sulfide crystals to grow erect. Then comes a coating of cadmium telluride and the top electrode, a copper and gold film. They attach the cell to a glass plate or make it flexible by pouring polymer solution on top and setting it.

"This is exciting progress in integrating engineered nanomaterials with a diversity of soft substrates for fabricating flexible and foldable high-efficiency solar cells," says Zhong Lin Wang, a materials-science and engineering professor at Georgia Tech. But the cell will have to compete with thin-film flexible solar cells made of silicon, cadmium telluride, and other materials, says Arthur Nozik, a physical chemist who studies nano solar cells at the National Renewable Energy Laboratory, in Golden, CO. As opposed to the new cell's flexibility, he says, "I think the selling point might be low cost."

For now, the researchers are exploring materials that could improve the cell efficiency. The top copper-gold layer, for instance, is only 50 percent transparent. If all the light falling on it went through, the cell's efficiency could already be double, Javey says. The researchers plan to make cells with transparent conducting materials such as indium oxide. "There is significant room for improvement, at least by two times, by simply improving or replacing our top contact material," he says.

The researchers also intend to try other semiconductor materials for the pillars and surrounding material. Javey says that the fabrication process is compatible with a wide range of semiconductors, and other combinations could up the efficiency.

Trying other semiconductor materials might also be important given cadmium's toxicity issues, Berkeley's Yang points out. Nevertheless, he says, "architecture is most important--materials we can continue working on. The beauty of this paper is the demonstration of how well the architecture works."