"This work demonstrates the importance of improving the performance of thin-film technologies," says Stephen Saylor, CEO of SiOnyx in Beverly, MA. SiOnyx is taking a different approach to increasing the absorption of red and infrared light in thin silicon devices. The company's black silicon material has a surface with nanoscale roughness that helps it absorb all visible and infrared light. The material's potential for solar cells has not been demonstrated yet.

Meanwhile, at the Ames Laboratory in Ames, IA, physicist Rana Biswas and his colleagues are using photonic crystals to make amorphous silicon solar cells more efficient. Their photonic crystal is composed of a lattice of tiny silicon cylinders inside an indium-tin-oxide layer. It could increase the efficiency of the solar cells by a maximum of 15 percent. But their amorphous silicon solar cells are only 0.5 micrometers thick, a tenth of the size of the MIT devices. "Generally, amorphous silicon-film solar cells need much less material, so cost goes down," Biswas says. "Plus they could be deposited on plastics. That's a big plus."

The MIT researchers aim to make thin-film silicon solar cells that are good enough to compete with conventional solar cells, Bermel says. By optimizing the photonic-crystal and grating structures, the researchers could squeeze the maximum efficiency out of the solar cells, increasing it to 13 percent. That would be comparable with the 13- to 15-percent efficiencies of some conventional solar cells.

The solar cells are far from practical right now. The researchers use an expensive technique called interference lithography to make the grating. Furthermore, the alternating layers in the reflector are deposited one by one, which is time-consuming. The researchers need to find a manufacturing technique that allows them to make the solar cells on a large scale and at low cost. "The ultimate question that must be answered is scalability," Saylor says. "To have a real impact, any solution must cost-effectively scale to mass production."

Bermel says that his team is already considering other production methods. One promising option is nanoimprint lithography, but they haven't tried it yet. "A 35 percent efficiency increase is clearly predicted in simulations," he says, "but the challenge is, 'Can you make it practically?' That's what we're working on."