THE figures certainly sound impressive. Last year, worldwide sales of solar cells, mostly made from crystalline silicon, grew by around a third. From Bolivia to Belgium, governments have started ladling out generous subsidies to help fuel this revolution. And now the next generation of solar cells are hitting the market – devices made from thin films of cadmium telluride that could generate even more power from your rooftop. By 2010, politicians promise, more than a million roofs across the US, Europe and Japan will be harvesting the power of the Sun.
Yet last November the industry received a major blow. BP Solar, the world’s second-largest solar cell producer, announced that it was closing its entire thin-film programme. Although BP’s thin-film cells work fine on the lab bench, it seemed that on the rooftop their efficiency would sometimes plummet. Instead, BP Solar plans to concentrate on old-fashioned crystalline silicon, the material presently used in 85 per cent of solar cells. Many had believed this stuff was due for retirement; now it looks set to be with us for a long time to come.
And therein lies the problem. Industry experts predict that the price of silicon cells will leap in the next two years. It isn’t that manufacturers are getting greedy or that subsidies are being withdrawn. It’s simply that the supply of silicon on which the industry relies is slowly being squeezed. Without some serious industrial innovation, the price hike this shortage will create could bring the entire solar energy revolution to a juddering halt.
Advertisement
There’s little doubt that it wouldn’t take much to kill the industry’s momentum. Despite last year’s spectacular growth, the solar energy business is still relatively small. Its worldwide generating capacity is just a few gigawatts, and although the cost of electricity from solar cells has halved in the last 20 years, a kilowatt-hour of electricity produced from sunlight still costs about 10 times as much as a kilowatt-hour produced from fossil fuels. If the industry’s rapid growth is to continue, the price of solar cells will have to fall further. But this in turn depends on continued help from subsidies – a financial lifeline that might be cut if governments lose confidence in the viability of solar power.
Though any number of exotic alternatives have been suggested, most solar cells are still made from bog-standard polycrystalline silicon, pretty much the same stuff that the electronics industry uses to make its chips (see “Power from the Sun”). In the past, solar cell manufacturers have mostly made do with offcuts from the electronics industry: the tops and tails of silicon ingots which contain more impurities than the bulk, for example, or wafers that do not meet the rigorous standards demanded by the chipmakers.
But since 1999, as demand for solar cells has risen, manufacturers have needed more silicon than these sources can supply. Their solution has been to supplement the offcuts with “electronics grade” silicon – the high-purity material suitable for microprocessors, with less than one impurity atom per billion silicon atoms. In the high-tech boom of the late 1990s, silicon wafer producers invested heavily in new capacity, but by 2000 the electronics industry was plunging into the most severe recession it has ever experienced. As a result, solar cell manufacturers were finding they could pick up the purer, electronics-grade silicon for little more than the $20 to $30 per kilogram they had been paying for offcuts.
Hard times for the chipmakers have provided the photovoltaic industry with some welcome breathing space. But this could be about to change. Demand for silicon, especially the cheaper offcuts, is set to outstrip supply. In 2001 the photovoltaic industry consumed about 5000 tonnes of polycrystalline silicon, of which about half was the purer electronics-grade material. Demand is projected to grow at about 15 per cent, and by 2010 the industry is expected to need almost 8000 tonnes of silicon, of which more than 5000 tonnes will have to be electronics grade (see Graphic). However, if the electronics industry bounces back and demand for microchips rises, the current surplus of electronics-grade silicon will dry up.
“At present there is no shortage because the semiconductor industry is in such a bad way,” says Hubert Aulich, managing director of British-based silicon manufacturer Crystalox. “Let’s assume it recovers in 2004; by that time the photovoltaics industry will be depending very heavily on electronics-grade silicon.” With chipmaking companies prepared to pay up to $70 per kilogram, the solar cell manufacturers’ costs will go through the roof.
Two studies, one carried out for the European Photovoltaic Industry Association and one for the European Union, have confirmed the bad news. They conclude that by 2006 the solar cell industry will be in dire need of its own supply of silicon. It’s certainly the most pressing issue in the EPIA’s strategic plan, says Aulich.
One answer would be for silicon wafer suppliers such as German-based Wacker-Chemie to dedicate more of their production to photovoltaics, producing lower-grade silicon faster and more cheaply than the material they make for the electronics industry. But why should they bother, especially if the electronics industry recovers?
Instead, many in the business hope that a manufacturer will step forward with a process designed specifically to churn out solar-cell-grade silicon at affordable prices. If silicon could be made for around $10 per kilogram, it could at last unlock the potential of the photovoltaics industry.
The reality is that making cheap polycrystalline silicon is far from easy. The conventional method starts with sand or quartz, both of which are mainly composed of silica – silicon dioxide. This is heated in a furnace and then reacted with hydrochloric acid to create trichlorosilane. This finally goes to a set-up known as a Siemens reactor, in which it passes across thin rods of pure silicon heated to 1100 °C. The high temperature decomposes the trichlorosilane into silicon, which deposits on the heated rods as pure, electronics-grade material.
However, all that heating eats up energy, and to make things more difficult, the silicon rods must be heated in three different ways. First the rods are warmed to 400 °C using external heaters. At this temperature the resistance of silicon drops to a point at which it is possible to heat the rods by passing an electric current directly through them. Once the rods reach 800 °C their electrical resistance drops still more and they can only be heated further by using a higher current. This multi-stage heating is difficult to control and requires two power supplies, an external heater and sensitive switching equipment. Since a lot of the heat is lost anyway, it is also extremely inefficient.
According to Aulich, companies such as Wacker Chemie could simply adapt this process to make cheaper, lower-grade silicon by increasing the deposition rate and doing away with the complex packaging steps required for silicon ingots destined for the chipmaking industry.
But a better alternative would be to extract silicon from silane (SiH4), which is produced from trichlorosilane using a catalyst, rather than from trichlorosilane itself. The main advantage of this route is that silane decomposes at more modest temperatures of around 800 °C. This means that less energy is needed than in the conventional process, so the silane route is far cheaper.
Keeping it cool
A number of companies are already developing processes that use silane. For example, Advanced Silicon Materials based near Butte in Montana has teamed up with the Renewable Energy Corporation from Høvik, Norway, to convert an electronics-grade silicon plant at Moses Lake in Montana to produce solar-cell-grade silicon from silane. The deposition will occur inside a conventional Siemens reactor, but at lower temperatures. Another American silicon wafer manufacturer, MEMC of St Peters, Missouri, has developed a process that deposits silicon on a fluidised bed of silicon spheres as silane is pumped through. This operates at temperatures as low as 700 °C.
But the most ambitious production process is being developed by two German companies, silicon wafer producer Deutsche Solar and chemical company Degussa, which have teamed up to form Joint Solar Silicon. In 1998 JSS bought key patents from the German chemical giant Bayer on a silicon deposition process that JSS claims has the potential to halve the price of the silicon for solar cells to the critical level of around $10 per kilogram.
Rather than using thin silicon rods, the JSS reactor uses a cylindrical reaction chamber made of silicon (see Diagram). The wall is heated to about 800 °C, and when silane is pumped in, it decomposes and deposits silicon onto the chamber wall. According to the patents, this design increases the surface area for deposition and improves the reaction’s efficiency. And better still, the surface area can be increased further by adding a silicon cylinder in the centre of the chamber so that the silicon layers forming on the walls of the cylinders eventually meet, creating a single block of the material.
This process offers significant advantages over the conventional Siemens reactor. In particular, the walls of the reaction chamber can be heated to operating temperature using only simple external electric heaters. “The Siemens process needs 100 kilowatt-hours of energy for each kilogram of silicon,” says Peter Woditsch, managing director of Deutsche Solar. “With our process, we think we can do this using half that energy.”
The key question is whether processes like this can get into production before the silicon shortage starts to bite. It usually takes at least four years to set up this kind of factory from scratch. But Frank Asbeck, chairman of Solar World, the parent company of Deutsche Solar, is confident that the JSS factory will be running by 2005 and will be producing 5000 tonnes of cheap silicon each year by 2007. JSS has already started building a pilot plant at the National Renewable Energy Laboratory in Golden, Colorado, and plans to build its first full-scale facility, with a capacity of 800 tonnes of silicon a year, next to the Degussa chemical plant in Antwerp, Belgium. The silane for the plant will be made by Degussa using a proprietary technique that Woditsch says can produce large quantities of the material cheaply.
However some competitors in the silicon industry remain doubtful of the claims being made for the JSS process. “It is seen by most in the industry as a dark horse,” says Iain Dorrity, marketing director of Crystalox. He even questions whether there will be a shortage at all: “Suppliers have a constant interest in talking the shortage issue up, customers in talking it down,” he says. “The truth probably lies somewhere in between.”
In the long run, one of the alternative solar technologies under development could make silicon solar cells obsolete. The thin-film amorphous silicon or cadmium telluride cells could yet come good, and newer technologies such as organic polymer films could also succeed. If one of these were to make a major advance, the entire issue of the silicon supply would become irrelevant.
But for the next 10 years at least, the prospects for solar power are tied closely to the availability of cheap and plentiful crystalline silicon. That’s certainly where firms like Solar World see the future. “Even if we go bankrupt because we bet on the wrong horse, so what?” Asbeck told Photon International magazine in an interview last year. “A tiny firm in Western Europe goes bankrupt and the world’s energy problem is solved. I think that’s a trade-off one could live with.”
Power from the Sun
Each year, the energy in sunlight falling on the Earth’s surface adds up to more than 10 times the amount tied up in all the planet’s reserves of coal, oil, gas and uranium. By 2050 renewable sources could be supplying half of the energy the world uses, and the most important of these could be solar cells.
At present almost 85 per cent of solar cells are made from polycrystalline silicon, although single crystals or thin layers of amorphous silicon are also used. To create a solar cell, atoms of other elements are used to dope the silicon. Then differently doped pieces of silicon are layered together, creating a diode. Free electrons in the silicon can only move one way across the junction.
To create a free electron, a photon of sunlight hitting the cell must have sufficient energy to knock an electron off a silicon atom. The one-way junction then causes the negatively charged electrons to build up on one side of the cell. Connecting the two sides of the cell into a circuit creates a flow of current, driven by the energy from the sunlight.
Polycrystalline silicon reaches energy-conversion efficiencies of up to 18 per cent. It only responds electrically to photons that fall within a particular range of energy levels, so the way to improve efficiency is to make cells that respond to a wider range of photon energies. To this end, researchers are trying to combine different semiconductors to optimise the energy range that solar cells collect. The most promising materials at present include cadmium telluride (with efficiencies of about 15 per cent), copper indium selenide (about 18 per cent) and certain organic polymers (up to about 4 per cent). Layers of these materials stacked on top of each other could collect more of the energy in sunlight. And by making the solar cells using thin films of material, production costs can be kept low.