"We now are very confident that we can meet the goals set by Darpa for its VHESC program," said Christiana Honsberg, an EE professor at the University of Delaware and co-principal investigator for the VHESC program. "After all, we were able to achieve 42.8 percent in just 18 months during the first phase."

That phase of the program, which just ended, consumed only $13 million of the $100 million promised by DARPA for all three phases. The next two phases, which will take 18 months each, will concentrate on optimizing the approach already begun, to squeeze out another 7.2 percent efficiency, for 50 percent total, while also reducing the cost of the design.

The consortium members are just as adamant about lowering the cost of the process to $1000 per square meter, but they admit that it is going take more than extending the architecture from five to six layers and optimizing the design, as it did for the 50-percent-efficiency goal. Today, gallium arsenide solar cells are only affordable for military and aerospace applications, and lowering the cost of solar cells using gallium, indium, nitride and silicon multiple dice will be a tall order.

"We have contributors with expertise in many different fields. The challenge will be to combine their various contributions to co-design a cost-effective solution," said Honsberg.

The novel technique used by the consortium involves multiple dice housed in hybrid packages with optical waveguides that split sunlight into its component wavelengths, focusing each band onto a solar-cell die optimized for just that wavelength. Multiple dice are required because the technique combines incompatible semiconductors, including single-crystal silicon, gallium arsenide, indium gallium arsenide and indium gallium nitride.

"Today, we have five different layers, using different semiconductor materials, and the final design calls for adding a sixth layer in order to achieve our goal of 50 percent efficiencies," said EE professor Allen Barnett, principal investigator for the VHESC program.

DuPont will be the principle industrial partner, along with others, for commercializing the hybrid solar-cell technology, although it will lean heavily on the University of Delaware's EEs to provide guidance in engineering a low-cost version of the current exotic design, which the consortium predicts will be in commercial production by 2010.

"Now we have to do all the tough engineering work needed to optimize the technologies and come up with a manufacturable design that will be commercially viable," said Honsberg.

The current design splits the spectrum, then concentrates the light with mirrors so that, by the time it hits the individual solar cells, it is more than 200 times stronger than normal, incident sunlight. Unfortunately, that approach requires expensive sun-tracking equipment that always keeps the cells pointed directly at the sun. For a commercial version, the concentration requirement will have to be re-engineered to hit below 20 times, so that conventional flat solar panels that do not have to track the sun can be fabricated.

Darpa's goal in funding the project, according to the agency, is to reduce the weight of the nearly 100-pound packs that soldiers carry today, by dropping the 20 pounds of batteries needed. Instead, soldiers will carry VHESC solar-cell arrays to charge radios, GPS units, night-vision goggles and other equipment during the day. The arrays will be 30 times smaller than today's solar-cell arrays, enabling soldiers to mount them on the tops of their packs, for mobile recharging, or on rooftops for stationary installations.

Industrial partners in the program include BP Solar, Blue Square Energy, Energy Focus, Emcore and SAIC. Research partners include the University of Delaware, the National Renewable Energy Laboratory, Georgia Institute of Technology, Purdue University, University of Rochester, Massachusetts Institute of Technology, University of California Santa Barbara, Optical Research Associates and the Australian National University.