Silane gas has killed and injured workers at cell-making plants. Can the photovoltaic industry live without it?
In 2007, outside Bangalore, India, an explosion
decapitated an industrial worker, hurling his body through a brick wall.
In 2005 a routine procedure at a manufacturing plant in Taiwan caused a
spontaneous explosion that killed a worker and ignited a blaze that ripped
through the factory, shutting down production for three months. Both incidents
shared a common cause—silane, a gas made up of silicon and hydrogen that
explodes on contact with air. And both incidents occurred in the same industry—solar
That critical factor means tractor–trailers traverse U.S. highways transporting tanks of silane as well as cargo ships laden with canisters of the gas cross the Pacific to factories in Asia, particularly China. It also means that a vast industrial structure in rural Washington State churns out solar-grade silicon, with silane as a by-product—plagued by shutdowns due to issues with the pyrophoric gas. A few thousand tons of SiH4 are produced annually worldwide, a number that will swell in the future as more and more microelectronics, flat panels and silicon photovoltaics are produced.Decreasing the thickness of the silicon layers—one answer to the cost challenge in photovoltaics—is not, however, a solution for the silane problem. If anything, amorphous silicon solar cells, which rely on relatively thin layers of silicon, employ more silane as part of their process, using the gas to deposit the thin layer of semiconducting materials that manufacturers such as Sharp and Uni-Solar need.
But there is an alternative. Quebec-based manufacturer SiXtron Advanced Materials has developed a way to make a more stable gas with similar properties via gasifying polymer pellets. The idea is to hook the SiXtron "Sunbox" to existing silicon photovoltaic manufacturing lines to provide a methyl silane gaseous mixture via the same pipes that would normally deliver silane from a canister. "The methyl silane gases are not pyrophoric, they are simply a flammable gas," says Bates Marshall, SiXtron's executive vice president of sales and marketing. "But a leak does not mean an explosion."
The key to the mixture's stability is replacing some of the hydrogen in typical silane with other molecules, such as carbon, to make the resulting gas less explosive. The silicon carbon nitride that this alternative deposits may even be better at helping reflect light back into a solar cell—and SiXtron's silane-free coating has been shown to reduce the loss in efficiency endemic to solar cells when they are first introduced to sunlight.
Georgia Tech's Rohatgi has been testing SiXtron's machine versus conventional silane solar cell production and found that they produce similar quality cells. "The cells we have made for SiXtron are just as good as what are being made in industry," he says. "It is great to have a technology which gives you gas without compromising efficiency. It is taking the risk out of the product."
Such an alternative may be vital, given that silane's behavior under various conditions remains a mystery, even to scientific experts. "Its behavior when released is unpredictable," admits Ngai, who has advised SiXtron. "As a result, silane has been involved in quite a number of significant incidents."
Already, such issues have driven major companies such as General Electric to opt for competing technologies, investing in thin-film solar cells made from cadmium telluride by PrimeStar. Amorphous silicon "is attractive because of the availability of silicon but you still have to have silane in high purity," says electrical engineer Danielle Merfeld, director of solar technologies at GE Global Research. Making solar cells from cadmium telluride "is a much more robust and forgiving process and material structure."
Regardless, silicon in its many forms makes up more than 80% of the current mix of photovoltaic production. It remains to be seen if the SiXtron alternative thrives, even though the company has partnered with solar manufacturing equipment provider Roth & Rau. Other potential alternatives in the past, such as disilane, faltered when they proved too expensive to make. "If industry sees it gives no performance loss and, if anything, it has potential for improving performance, then I see no real reason for industry not to embrace it," Rohatgi says. "It will take time."