ENERGY

BP Solar Ditches Thin-Film Photovoltaics

A big setback to industry's vision

Thin-film photovoltaics—with micrometer-thin semiconductor coatings on glass or metal that convert sunlight into electricity—have been poised to revolutionize solar energy for two decades. They require much less material per unit electricity yielded than conventional photovoltaics made from semiconductor-grade crystalline silicon wafers, and are also far simpler to produce.

But in November that vision of thin films soon displacing crystalline photovoltaics blew a fuse. One of the world's largest producers of photovoltaic solar cells, BP Solar (Lithicum, Md.), abruptly announced plans to cease production at plants making thin-film cells from amorphous silicon (a-Si) and cadmium telluride (CdTe). The CdTe composite is a leading contender for next-generation thin-film photovoltaics, though its toxicity is a serious drawback, and a-Si is the most commercially advanced thin-film material. Overnight, all mention of thin-film technology vanished from BP Solar's Web site, erased even from the corporate history page that had proudly chronicled two decades of research and development.

"You can't say this is good news. This is a blow to thin films," says Larry Kazmerski, who runs the solar program at the U.S. Department of Energy's National Renewable Energy Laboratory (NREL, Golden, Colo.). The lab has been spending US $50 million each year helping manufacturers, including BP Solar and the Royal Dutch/Shell Group subsidiary Shell Solar (Amsterdam), scale up production of thin-film modules.

The announcement has been all the more shocking because the overall outlook for the photovoltaics industry has seemed to be brightening markedly. Sales of photovoltaic modules grew by 34 percent in 2001, and healthy growth was expected last year, too—15-;20 percent was the projection from the research firm Strategies Unlimited (Mountain View, Calif.). Increasingly, the modules are powering not just remote villages and road signs, but homes and businesses connected to the power grid in bustling economies.

Yet growth was from a microscopic base. Photovoltaic modules made using semiconductor techniques generate electricity at three to four times the direct cost of power from coal, gas, or nuclear plants. That is precisely why NREL, BP Solar, and other technology developers have been pouring hundreds of millions of dollars into development of thin-film photovoltaics, seen as the best hope of slashing module costs and making solar power competitive on the grid.

Just a mirage?
Unfortunately, thin-film technology has fared better in the lab than in the market. Low efficiency has dogged a-Si thin films: commercial modules convert a paltry5-;7 percent of incident solar energy into electricity, about half as much as crystalline silicon modules, so twice as many panels are needed to achieve a desired output. Meanwhile, advanced, more efficient thin films have been beset with production and design problems.

BP's CdTe-based Apollo thin-film technology was a case in point. In 1998, to great fanfare, Vice President Al Gore cut the ribbon at BP's first Apollo plant in Fairfield, Calif. But production at the factory—designed to make enough modules each year to generate 8 MW—has never amounted to more than 1 MW annually. Creating films to exacting specifications over a square meter of glass proved harder than expected. And the first commercial modules, shipped early this year, fizzled after a few weeks on rooftops as their efficiency slipped from 8 percent to 6 percent.

The Apollo production problem was the last straw for at least one important customer, PowerLight Corp., a solar power system integrator in Berkeley, Calif., and North America's largest buyer of photovoltaic modules. "We're not going to use any more of this material," says PowerLight President Daniel Shugar. "That's it."

Shugar notes, though, that Power-Light will continue to buy thin-film modules made from amorphous silicon on steel, which are flexible and useful in certain niche applications. And he is keeping the faith as far as the photovoltaics industry's general future is concerned. In essence, as he sees it, even as the outlook for second-generation thin films has deteriorated, prospects for current-generation crystalline silicon cells have improved because of a steep drop in module cost, thanks to large-scale robotic production, an order-of-magnitude decrease in the price of silicon ingots over the past 15 years, and sawing methods that generate thinner wafers.

Standing by thin films
NREL's latest survey of manufacturers provides considerable support for Shugar's view of things. Crystalline modules cost just $1.72/W to produce last year, down from $4.23/W a decade ago and well below the $2.40/W average cost for thin-film modules. NREL projects that costs of both technologies will ratchet down to roughly $1/W by 2007.

That last year is a significant date for BP Solar, too, because its parent, the oil and gas giant BP (formerly British Petroleum), had promised to derive $1 billion in revenue from photovoltaics by then. BP Solar realized that investing more in thin films was not the best way to get there. "While [thin-film] technology continues to show promise, lack of demand for the material and present economics do not allow for continued investment," says the company's president and CEO, Harry Shimp. Instead, he says, BP will make more crystalline and polycrystalline silicon modules, which accounted for about 90 percent of its estimated $300 million in sales last year.

For its part, NREL is standing by thin films. Kazmerski says that BP's decision only reinforces the need for federal R&D investment. "We have to ensure that we can take new technologies from the lab bench to manufacturing faster and with more confidence," he says.

NREL will continue supporting scale-up of module production by its remaining partners. First Solar (Phoenix, Ariz.) is carrying on with its own CdTe production efforts, and several other firms—notably Shell Solar, formerly Siemens' solar division, continues to work on copper-indium-diselenide thin films. Kazmerski predicts that thin films will achieve economies of scale and take off in the market by 2010. He even thinks they could rival the efficiency of crystalline. After all, NREL recently confirmed thin-film lab cells with efficiencies above 19 percent.

Shimp says that BP will continue R&D on thin films, and expects that they will eventually have their day in the sun. He agrees that the key is boosting efficiency—that is, efficiency of commercial thin-film modules. "If someone can get efficiency in the low teens, thin films will have a shot."

—Peter Fairley

PHOTOS: CLAY MCLACHAN/GETTY