Solar Market Powers Silicon

High profits and low risk.That is the sweet spot that polysilicon producers find themselves in these days now that their product, key to the manufacture of photovoltaic solar panels, is in desperately short supply.

In July, for example, the Chinese photovoltaic cell maker Suntech committed to buy at least $5 billion worth of polysilicon over 10 years from St. Peters, Mo.-based silicon wafer producer MEMC. Suntech also agreed to make interest-free loans to MEMC to help it build new plants needed to fulfill the contract. The deal is not unusual.

Solar energy is a relatively new market for polysilicon manufacturers. They traditionally made polysilicon, known more formally as polycrystalline silicon, mostly for the semiconductor industry, which uses it to produce silicon wafers. They sold solar cell companies either surplus polysilicon or off-spec material that was not sufficiently pure for semiconductor fabrication.

In recent years, demand from solar cell manufacturers has grown to a point where it outstrips the ability of polysilicon manufacturers to supply them. Industry executives are increasingly optimistic about the prospects for the solar industry. "Growth has been averaging 35-40% annually for more than five years," says Richard S. Doornbos, president and chief executive officer of Hemlock Semiconductor, the world's largest producer of polysilicon.

"A lot of analysts believe, and I do too, that there are some pretty strong fundamental drivers for the growth of solar industry," Doornbos adds. Like others, he expects that within two years, the solar industry will consume more polysilicon than the semiconductor industry.

The large and growing requirements of the solar industry are giving rise to a wave of capacity expansion announcements by polysilicon companies this year. According to the stock brokerage Piper Jaffray, the solar industry will consume more than 30,000 metric tons of polysilicon in 2008, compared with 27,000 for the semiconductor industry. Last year, the solar industry used 17,500 metric tons while the semiconductor sector used 21,500.

Hemlock is spending $400 million to $500 million to expand polysilicon capacity at its main site in Hemlock, Mich. The firm, a joint venture of Dow Corning, Shin-Etsu Chemical, and Mitsubishi Materials, is also undertaking a search for a new site somewhere else in the world in order to set up a second plant within five years.

Wacker Chemie, the world's number two producer of polysilicon, announced this summer that it would build a 4,500-metric-ton-per-year plant in Burghausen, Germany, to supply solar cell manufacturers. Also in Germany, Deutsche Solar, the solar wafer subsidiary of solar cell manufacturer SolarWorld, is teaming up with Degussa to bring on-line an 850-metric-ton polysilicon plant that makes use of a new process.

Most of the new facilities under construction will come on-line in 2008 or 2009. In the meantime, manufacturers of solar cells have to manage the shortages as best they can.

Peter Woditsch, CEO of Deutsche Solar, tells C&EN that it's obvious why polysilicon supply has fallen so far behind demand: the cautiousness of the companies that make it.

Spurred by bullish forecasts provided by semiconductor manufacturers during the Internet boom of the late 1990s, polysilicon companies made significant investments in expanding their production capacities. But the semiconductor boom turned to bust in 2001, causing an oversupply of polysilicon. This experience of not so many years ago has made silicon producers conservative about investing.

But the risks involved in expanding capacity appear to be particularly minimal these days. Polysilicon producers have the luxury of starting construction on new plants after having signed supply contracts that commit most of the output for several years.

"The commodity suppliers realize that they have a pretty good situation at hand, and they take advantage of it in a mostly constructive way," says Julie Blunden, vice president of external affairs at SunPower, a major manufacturer of solar cells based in California.

Earlier this year, SunPower signed a contract with DC Chemical, a diversified chemical producer based in South Korea. Under the agreement, SunPower will buy at least $250 million worth of polysilicon over four years and will prepay for the shipments to help DC finance the construction of a new plant that is expected to open in 2008.

Surprisingly, DC has never produced polysilicon, and this does not worry SunPower. "We're not a chemical company, but they are, and there is a lot of expertise associated with being a chemical company," Blunden says, noting that DC already produces the precursor gases needed to make polysilicon.

Blunden says the deal with DC adds another supplier to the list of companies already supplying polysilicon to SunPower. In the spring, the California firm increased to $500 million and extended by two years an existing supply agreement with Japan's M.Setek, an established polysilicon producer.

Jesse W. Pichel, a senior analyst who covers the semiconductor and solar power industries for Piper Jaffray, believes that polysilicon producers are enjoying huge profit margins these days. He estimates that it costs about $30 to produce a kilogram of solar-grade polysilicon at present. The contract sale price, meanwhile, is $70 per kg on average, "if you're lucky to have a contract," he says.

As for the spot market price, it's as much as $200 per kg. Pichel notes that there are many producers of solar cells in China and that they meet about half of their polysilicon requirements from the spot market.

Hemlock's Doornbos cautions that margin estimates made by outside organizations are not reliable "because it's hard to get the information."

Pichel expects the tight supply conditions to prevail for some time. He estimates that as much as 98% of the polysilicon that will be produced in 2007 has already been allocated to buyers under forward contracts and that 95% of 2008's polysilicon is also preallocated. He says no manufacturer except MEMC has any significant amount of capacity to spare, and he expects MEMC to soon supplement its Suntech deal with a supply agreement with another major company.

Owing to the huge margins and strong demand forecasts, many firms are trying to set themselves up as manufacturers of polysilicon. Besides major chemical companies such as DC Chemical, smaller firms in China are also trying their hands at producing the material. Hemlock's Doornbos contends that new entrants will face some real challenges in mastering the processes involved. "We're talking about making the world's purest materials," he says. "One should not underestimate the technology challenges."

Making polysilicon is "quite different from the typical chemical industry processing," Doornbos adds. He expects that several of the new entrants will ultimately become successful players but doubts this will happen quickly.

Polysilicon is cumbersome to make. The raw material is metallurgical-grade silicon, obtained from the reduction of mined silicon dioxide. This silicon is then reacted with hydrogen chloride to form trichlorosilane.

A Wacker spokesman tells C&EN that in the traditional Siemens process, trichlorosilane and hydrogen gas are sent into a reactor containing thin silicon rods. The rods are heated to 1,000 °C, causing the trichlorosilane to decompose and deposit itself as polycrystalline silicon.

Semiconductor wafer manufacturers and some solar cell manufacturers require monocrystalline silicon. This higher grade material is obtained by crushing polysilicon, melting it, and then growing it into an ingot with a singular crystal orientation. Boron or phosphorus can be added to improve connectivity.

Because they are sold out of polysilicon at present, major players such as Wacker and Hemlock do not compete for customers. But normally, Doornbos says, competitive differences are found in the refinements to the Siemens process that each firm has come up with over the years. These refinements allow better control of the production process and the quality of the silicon produced. He adds that the reputation each firm has developed over the years is another differentiating factor between companies.

And if there weren't a shortage, says Piper Jaffray's Pichel, the main basis of competition would be each firm's cost of production. "Whoever has the lowest cost of making polysilicon wins," he says.

The booming demand for solar cells worldwide is the result of many factors. Rising prices for fossil fuels and concerns over their availability, Doornbos says, are spurring interest in solar energy. In addition, nations that signed the Kyoto protocol are looking for ways to reduce their emissions of greenhouse gases.

Deutsche Solar's Woditch points to the regime of subsidies pioneered by Germany to encourage households and businesses to invest in solar energy generation. The subsidies are the result of a collective desire to protect "the Earth we borrowed from our children," he says. The subsidies are necessary, he maintains, because at present, "the cost of electricity produced by a photovoltatic process is not competitive against electricity produced by depreciated power stations."

The growing list of U.S. states and countries that subsidize solar energy in a similar way to Germany now includes California, New Jersey, France, Greece, Italy, Portugal, South Korea, and Spain.

Woditch adds that the solar industry's relative immaturity is one reason the cost of producing solar energy is so high. "It is not well-developed, it is not large-scale, and production is not automated. It's a handicraft," he says.

The polysilicon shortage is prompting companies to look at new ways to make solar cells. Whereas last year the solar cell industry grew 30% as measured in terms of new wattage produced, growth this year will be only 5%. Until last year, Pichel says, solar cell manufacturers could still get their polysilicon from the stockpiles of material accumulated in the industry glut of 2001-02.

In particular, thin-film technology is gathering steam, Pichel says. Its prime advantages are that it allows solar cells to be produced at a lower cost and that it does not consume any polysilicon. Instead, the key material on the surface of the cells is amorphous silicon, which is obtained from silane gas, or an alloy like copper indium gallium. Thin-film solar cells are made by U.S. companies Energy Conversion Devices and First Solar, as well as a handful of companies in Germany. Thin films already represent 5% of the solar energy market, Pichel estimates.

Dow Corning announced in early September that it had developed a grade of metallurgical silicon that could be blended with traditional polysilicon to make solar cells. The company said the availability of the material could help relieve polysilicon supply shortages. Dow Corning CEO Stephanie Burns tells C&EN that commercialization of solar-grade metallurgical silicon is an important achievement that Dow Corning had been working toward for a number of years.

Opinions are divided on how widely Dow Corning's metallurgical silicon will be accepted. At SunPower, Blunden says the availability of solar-grade metallurgical silicon is a welcome development that will help reduce supply pressures in the industry. But she says SunPower will stick with its monocrystalline silicon process and has no intention of using metallurgical silicon.

Piper Jaffray's Pichel does not see the point of using metallurgical silicon in a solar cell, calling it far less efficient at generating electricity than polysilicon. He says the efficiency of a cell decreases in proportion to the amount of metallurgical silicon that is added to it. "The last thing you want to do if you're a solar cell producer is to lower the efficiency of your cell," he says.

Even though new materials may be promising, it will take many years for them to become a credible alternative to polysilicon. As a result, shortages of polysilicon will continue until at least 2008 when new facilities come on-line. And if the cost of producing solar-based electricity ever becomes competitive with traditional means, Pichel predicts that "there will never be enough polysilicon to go around."