Thin-film technology has propelled one supplier past silicon solar cell makers. But will the rarity of the raw materials bring an early end for the upstarts of PV?
Cyrus Wadia, scientist at Lawrence Berkeley Laboratories in California and an adviser to President Barack Obama's administration, has a dream: to get solar energy into the hands of seven billion people. And he is pretty sure that silicon is not going to do it.
Silicon has one thing going for it: there is a lot of it. 'When you look at the starting point, sand, you think: 'Wow, it's got to be really abundant; it's got to be cheap',' says Wadia.
The problems for silicon start once you start purifying it. 'You have to put in a lot of energy to break that silicon-oxygen bond. It's a very strong bond,' Wadia explains, comparing the energy needed to reduce the iron in ferrous oxide to its metallic state. 'It is an order of magnitude lower.'
The cost of silicon extraction and purification is exacerbated by how thick the material needs to be to capture light efficiently. 'Silicon's absorption coefficient is relatively low compared to some of its colleagues, so the thickness has to go up,' Wadia explains. 'But as the thickness increases, the chance of an electron or hole getting out goes down. To increase the chance of a charge carrier making it out of the solar cell, you have to make the silicon more pure.'
This means that metallurgical-grade silicon produced in a basic arc furnace is not good enough. The silicon has to go through another phase of heat-assisted purification to make it to chemical grade. Studies have found that half or more of the total energy usage needed to make crystalline-silicon solar modules comes from the purification stages alone. Wadia says there are tricks that may make it possible to use the initial metallurgical-grade silicon for energy generation, but they do not exist yet.
'Because we have to make it thick, it limits our form factor and limits our creativity of how to package it,' says Wadia, adding that silicon's energy cost does not make it a strong contender for a photovoltaic technology that can be deployed in both the developed and the developing world, despite its dominant market share today.
Wadia argues that the evolution of the solar cell has favoured silicon so far but that growing demand for solar generation will force the industry to examine other options.
The main alternative right now is to go to thin-film technologies that are made using much more exotic raw materials than sand. Because they are much thinner, they have the advantage of using far less in the way of expensive raw materials.
One of the thin-film options has proved so successful that it has helped establish First Solar as the world's largest producer of photovoltaics: surpassing any of the established silicon cell and panel manufacturers, such as Suntech and Sharp. But as the sole high-capacity thin-film producer, crystalline silicon as a technology still dominates the market with close to 80 per cent market share.
First Solar has built its dominant position on the back of two exotic metals: cadmium and tellurium, often shortened to 'cadtel' by people in the industry as the symbols CdTe do not make for an easy-to-say abbreviation. It is far from the only option. A growing number of manufacturers are exploring the material systems around a mixture of copper, indium and gallium with either sulphur or selenium under the acronym CIGS. Of CIGS, Otte says: 'There are so many emerging companies in this segment.'
CIGS has even enticed Taiwan Semiconductor Manufacturing (TSMC), the world's third largest maker of silicon chips. The company has licensed technology from startup Stion to give it a head start in production for a $260m fab that TSMC expects to have a capacity of 200MW per year by the end of 2012. The company says it will spend more for a greatly expanded second phase - turning out 700MW per year when the second phase is complete.
Professor Ayodhya Tiwari of Swiss federal laboratory Empai says the efficiency of CIGS has reached 20 per cent, in individual cells at least, with CdTe following at 16.5 per cent. What they lack in efficiency compared to the best crystalline silicon cells, which have demonstrated conversion efficiencies of more than 24 per cent, they potentially make up for in terms of manufacturing cost. Relying on films just a few micrometres thick, they need much less in the way of high-purity source materials than cells that rely on the production of thick crystalline silicon wafers.
The cost advantage of thin-film has allowed First Solar to power its way to the top of the PV market. Silicon-cell producer Q-Cells has joined the thin-film club, claiming an efficiency of 13 per cent for its CIGS-based modules in June, the highest recorded for commercial rather than laboratory-produced cells.
Tiwari argues that, even though First Solar has built the dominant position in CdTe thin-film cells, there is still room in the market for other entrants to at least carve out a niche - ranging from his dream of a 'solar pen', which has a roll-out thin-film PV foil designed to charge mobile devices, to larger-scale designs. 'This metal system is very robust and there are similar metal systems. There are opportunities for other companies to develop this technology,' Tiwari says.
Wadia argues that, despite their rapid progress, these exotic materials are not good long-term choices. Volume production will ultimately become their Achilles heel because there simply is not enough readily extractable material to go around.
Indium, which originally was extracted as a byproduct of zinc mining, saw its price increase more than five times during the past decade, according to the US Geological Survey, because of its use in flat-panel displays. World production increased during that time by just 20 per cent. Indium for solar panels will have to compete with customers in the display market for the foreseeable future, although efforts are underway to find more abundant alternatives to indium tin oxide, today's most viable transparent conducting material (E&T, Vol 5 #9).
The problem for CIGS is that competitive pressures between industries can cause sudden spikes in price. An example was the sudden boom in the price of palladium at the end of the 1990s as electronics manufacturing surged and carmakers decided to swap from platinum to palladium for catalytic converters in response to a rise in platinum prices. Companies making capacitors suddenly found the palladium they needed for contacts in short supply.
As CdTe production has ramped up, tellurium prices followed. In 2000, tellurium cost just $35,000 per tonne. By 2008, one tonne of the metal was worth more than $210,000 and production has not increased to accommodate tellurium's new role.
World reserves of tellurium, assuming that the primary source is as a byproduct of copper refining, were estimated by the USGS at around 22,000 tonnes. Turning to other sources should raise that total but by how much is not yet clear. First Solar uses around 100 tonnes of tellurium to make 1GWp of solar capacity, with production ramping to 1.8GW per year by 2012. A calculation made by Wadia and colleagues estimated that the comparatively small reserves will limit CdTe's usage to no more than the US electricity consumption - a much higher figure than the installed solar base but still a long way short of Wadia's target. The situation is not much better for CIGS because of the relative scarcity of indium.
Wadia insists more work needs to be done on more abundant materials that do not suffer the punishing energy cost that silicon production has. One material system that Wadia favours is fool's gold: ferrous sulphide. Based on two abundant materials that are easy to extract - using far less energy than that needed for silicon - FeS has a lot going for it. Everything, except a lot of experience in photovoltaic production. He claims the problem is that people have not focused on FeS or similar systems. Many of those working in solar have, instead of exotic metals, looked at organic options, such as nature's own offering, chlorophyll.
Karl Leo of the Institute of Photonics and Microsystems at the Technical University of Dresden is placing his bets on organic with silicon in second place because of its abundancy. 'Cadmium telluride is probably not viable in the long term.'
Total's Vermeersch agrees with Leo that organic technology will win through because the raw materials are far more abundant and also 'to me, it's the only really flexible technology' - as one that can be wrapped around objects as narrow as pens.
Tom Tibbits, director of product development at Quantasol, says concentrator systems based on semiconductors such as gallium and arsenide are more scalable because they use comparatively little in the way of photosensitive material. Lenses concentrate sunlight onto the cells, potentially leading to very high conversion efficiencies that could reach as high as 50 per cent. However, this technology is only suitable for large installations today because the panels have to actively track the Sun.
CdTe and CIGS backers point out that the mass-market dream of the solar future proposed by Wadia is still a long way off: thin-film makers have years of growth to work through before they have to worry about running out of raw materials or switching to a different technology. By that time, organic technology could have progressed to the point where metallic systems are not worth the investment, or systems that combine rarer materials with steerable lenses in concentrators become cheaper to make, eking out their precious constituents.