Sun and sand: dirty silicon could supply solar power.

Scientists have proposed a way to control the distribution of contaminants in silicon, potentially opening up the use of cheaper, "dirtier" starting materials for making solar cells. In a study published in the September Nature Materials, the researchers predict that the strategy could lower production costs of solar cells.

Silicon is the second most abundant element in Earth's crust, but nature's primary sources of silicon--sand and quartz--are tainted with metals. Converting silicon from these sources into superpure crystals is an expensive and time-consuming process.

While there had been enough pure stock for the electronics industry, the needs of the growing photovoltaic industry--which uses silicon for more than 90 percent of its solar cells--caused overall demand to exceed supply in 2004, notes Eicke R. Weber, a materials scientist at the University of California, Berkeley and the Lawrence Berkeley (Calif.) National Laboratory. This triggered a drastic price increase in pure silicon, dealing a blow to the solar cell makers.

Silicon stock that is less pure and therefore less expensive is available, says Weber. But the increased amounts of iron, copper, and other metal contaminants in these stocks reduce solar cell efficiency. Clusters of these metal atoms attract the solar cell's charge-carrying electrons, reducing the amount of current that the cell can generate.

Weber and his colleagues set out to see whether they could minimize the toll taken by these clusters without having to get rid of them.

To do this, they turned to Lawrence Berkeley's synchrotron, a circular accelerator approximately 65 meters in diameter. The machine generates X rays intense enough to identify within silicon samples individual metal dusters on the order of tens of nanometers in diameter. Weber's team mapped the distribution of the clusters and used a sophisticated technique for measuring how far charges traveled in the samples, an indicator of the material's efficiency in converting sunlight into electricity.

The researchers found that silicon hosting larger but fewer numbers of clusters performed better than did samples with smaller but many more clusters. They tested this result by heating samples and then cooling them at different rates, which enabled the researchers to control the distribution of the metal. Weber's team found that silicon with micrometer-size clusters, spaced hundreds of micrometers apart, was four times as efficient as silicon with more-finely-distributed, nanosize clusters.

"Without changing the total metal content--only changing the way it is distributed--we get a drastic change in the electrical property" says Weber. "If it is possible to concentrate the metals in a few big clusters, in principle, one can make good solar cells out of ditty starting material."

"It's excellent work," says Bhushan Sopori, an electrical engineer at the Department of Energy's National Renewable Energy Laboratory in Golden, Colo. But he cautions that "you often do not have as much control [over metal impurities] as you think" when growing silicon crystals.