Solar industry creates rocketing demand for silicon

Craig Morris

At Intersolar, Europe's largest trade fair for the solar industry, the main topic in the past few years has been "grey gold" — and how to get it.

The solar industry is currently booming at rates with which the IT sector was once familiar: double-digit growth is nothing unusual. Indeed, in the past dozen years the global solar industry has grown by a full 30 to 40 per cent annually. In 2005, it even exceeded 44 per cent growth; in 2006, again by around a third. (Note that the figures cited in the press vary slightly depending on whether reference is made to installed capacity or cell production.) Germany remains the driving force with more than half of the world market in terms of installed capacity, but Spain dominated the global market in 2007 (official figures have at the time of writing not been released) after growing by 200 per cent in 2006. In 2005, solar power production in Germany grew by 152 per cent. Not surprisingly, Intersolar is moving from its historic venue in the relatively small town of Freiburg, Germany (pop.: 220,000) to the big city of Munich in 2008, where the fair expects to offer a full 80 per cent more floor space. The parking lot on the Freiburg Fairgrounds was already covered with makeshift tents in 2006, with the fair maxed out at around 30,000 square metres of floor space.

Growth would be even faster if a bottleneck in the supply of solar-grade silicon had not been slowing down production since 2004. Experts estimate that production lines for solar panels in Germany were only running at around 60 per cent capacity in 2006, a severe drop from 80 to 90 per cent in the previous year. But there are now signs that production has been ramped up enough to meet demand.

For many years, the solar sector bought waste silicon from semi-conductor production. As solar-grade silicon does not have to be as pure as electronics-grade (99.99 per cent instead of 99.9999 per cent for semi-conductors), the photovoltaics industry used to be able to make do with such relatively inexpensive waste products. While photovoltaics still only makes up some 0.3 per cent of power production in Germany, relatively large amounts of silicon are required for solar cells – some 13 tonnes per megawatt-peak.

The price of a kilogram of solar silicon doubled from 2004 to 2006 to a full $60 US at the beginning of 2006 in long-term purchase agreements. Those who had not entered into such agreements fared much worse, with prices on the spot markets occasionally exceeding $100. It was an especially enormous increase if we remember that prices had fallen down to close to $20 back in 2000 during the dot.com crisis. The new millennium began with silicon producers complaining about overcapacity, manufacturing at a loss, and putting plans for the construction of new production plants on the back burner. A look at the shipping statistics published by SEMI shows that the market collapsed by about a third in 2001.

Only a few years later, prices were rising faster than production figures. The market for polycrystalline silicon grew by an estimated 45 per cent in 2005 from $825 million to $1,200 million, though production volumes only grew by 25 per cent from some 25,000 tonnes to 30,000 tonnes. In 2006, production increased to around 34,000 tonnes. In 2006, the photovoltaics industry made up a third of that at around 13,000 tonnes after only 10,000 tonnes in the previous year. In other words, most of the growth in 2006 was gobbled up by the photovoltaics sector. In fact, by 2009/2010 demand for solar silicon could exceed demand for electronics-grade silicon.

Not surprisingly, the silicon industry is shifting its priorities. This year, the world's largest manufacturer of polycrystalline silicon, Hemlock Semiconductor, may soon be surpassed by Germany's Wacker-Chemie in terms of production even though both are ramping up output dramatically. In May of 2006, Norway's Renewable Energy Corp. (REC) became the world's second largest wafer manufacturer when it went public along with its subsidiaries SGS and ASiMi. Japan's Tokuyama Corp. has itself fallen back into fourth place.

Solar panel manufacturers are themselves having to invest in the production of silicon. Germany's SolarWorld AG was the first to see this trend coming. The company, which went public in 1999, got into the production of silicon way back in 2000, when it took over Bayer Solar GmbH of Freiberg in Saxony (not to be confused with the Freiburg in southwest Germany). Today, Solar World operates a large recycling facility for old solar cells in Freiberg. In addition, SolarWorld has also entered into long-term contracts both with Hemlock and Wacker to ensure a supply of solar silicon.

The market has thus begun to clamp down on the small producers of panels, such as Freiburg's Solar-Fabrik AG. Nowadays, it is almost impossible to stay at the end of the supply chain and only produce panels, which is why the Freiburg company expanded downstream with the revenues it received from its IPO in 2002. Like its big brother and competitor SolarWorld, Solar-Fabrik now also manufactures some of its own wafers into solar cells, which are then used to manufacture solar panels.

The entire industry is working feverishly to make cells thinner in order to make do with less silicon per cell. Some of the companies are also looking for alternatives to silicon. While silicon wafers make up more than 90 per cent of the market today, thin-film technology can make do with only a 50th as much silicon.

Back when oil multinational Shell sold its crystalline solar division to SolarWorld in February of 2006, it announced that it would be devoting all of its efforts to thin-film solar production. While the takeover made SolarWorld number one on the US solar market overnight, Shell's departure from crystalline photovoltaics spoke volumes. At the same time, thin-film photovoltaics has a much lower efficiency than polycrystalline and monocrystalline cells at about seven to 10 per cent under field conditions, compared to a full 15 to 16 per cent for poly- and mono-crystalline, respectively. But thin-film is cheaper, and rising silicon prices could give thin-film photovoltaics the upper hand; after all, what matters is generally not the efficiency of a given surface area of photovoltaics (aside from outer-space applications), but rather the cost per peak kilowatt-hour.

Other combinations of metals are also being heavily researched in addition to organic cells based on pigments. While the latter still have unacceptably short service lives, with rates of efficiency plummeting within weeks if not days, the former simply involve the use of other metals that may not be in great demand now, but could soon be only on sale at a premium. As the market experts at Photon Magazine recently pointed out, we will probably never run out of silicon, but if the world has to rely on relatively rare metals for photovoltaics, then PV may never make a significant contribution to our energy supply.

For the IT sector, the question is whether and when demand for solar-grade silicon will ever raise prices for electronics-grade silicon. Sematech already voiced its concern back in 2005.

(Craig Morris)