Solar cell converts water into hydrogen.

Some people envision a future world powered only by the sun and water. No longer dependent on petroleum, this society would harness the sun's energy to produce hydrogen gas, a clean-burning, renewable fuel that can be packaged in fuel cells or piped directly to homes.

Now, John A. Turner and Oscar Khaselev of the National Renewable Energy Laboratory (NREL) in Golden, Colo., have taken a small step toward that lofty goal. They have fabricated a solar cell that harnesses sunlight to produce hydrogen gas from water. This photocell absorbs light and converts it into an electric voltage strong enough to split water into hydrogen and oxygen.

Other systems that produce hydrogen from water follow the same principle, says Turner, but they keep the light-absorbing and water-splitting components separate. The new photocell combines the two parts into "a single, monolithic device," he explains.

Eliminating the need to channel the absorbed energy from one component to the other boosts the device's efficiency to 12.4 percent, nearly twice as high as other methods have achieved. Improving efficiency is important for making such systems commercially viable.

The photocell consists of a layer of gallium indium phosphide, a semiconductor that absorbs visible light, laid upon a double layer of gallium arsenide, which absorbs infrared light. Both materials convert light into electric voltage.

To complete their setup, the researchers immerse the photocell and a platinum electrode in a weak acid solution. When they shine a light with 11 times the intensity of sunlight on the photocell, the researchers see hydrogen bubbling off the surface of the cell and oxygen rising from the platinum electrode.

The team reports its findings in the April 17 Science.

Although the integration of components in the photocell may increase its efficiency, that approach adds "a heck of a lot of complexity," says Allen J. Bard, an electrochemist at the University of Texas at Austin. The need to immerse the whole chip in a solution could be a disadvantage, he adds.

The researchers did see some damage on the photocell surface as the hydrogen bubbles collected. Although the cell operated for 20 hours without a problem, Turner says, a practical device must remain stable for 5 to 20 years.

Turner thinks this particular device has reached its highest possible efficiency but says a different combination of materials might do better. Scientists have predicted an 18 to 24 percent theoretical efficiency for these kinds of systems, he notes.

At present, hydrogen is much too expensive to compete with oil, and the NREL device doesn't purport to decrease the cost, Turner stresses.

A crossover to hydrogen fuel might not be possible for 50 to 100 years, says Bard, but "no question, eventually we're going to run out of fossil fuels." Solar conversion of water into hydrogen is "a viable alternative," he adds, but "the world is dreaming of a very simple, low-cost system. No one's found it yet."