The photovoltaic trade website Solarbuzz reports that the latest wholesale price of a solar panel module is $4.85 per watt. Four years ago, in August 2003, it was $4.62. Why no progress? Isn’t silicon abundant, lying round the Sahara in million-ton sand dunes? And don’t solar cells come from the same glamorous technology that delivers Moore’s law year after year?
These are the wrong grades of silicon. Rocks and sand are largely made up of silicate compounds, which have to be refined into elemental silicon. From then on, it’s a question of purity. The world market for elemental silicon looks roughly like this (from various unofficial web sources, don’t bet the farm on the numbers):
|Annual demand, tons||900,000||6,000 (2003)||20,000|
|Growth rate of demand, annual||low||20-25%||low|
|Manufacturing process||arc furnace||as semiconductor||vapour deposition|
|Acceptable impurities||<1%||<10 ppm||<0.1 ppbn|
|Price per kg.||$1-2||<=$50||$50|
The problem for the PV customers for silicon is that they are a fast grower sandwiched between two mature sectors growing roughly in line with the economy. Bulk silicon is used in old-economy alloys and sealants; and while demand for semiconductors grows rapidly in value, their extra capability is crammed onto roughly the same physical volume of raw material. Unfortunately there is no appropriate process for making PV-grade feedstock. Metallurgical-grade silicon, smelted by simple Victorian technology, largely in China, is cheap but too impure to work in solar cells. So you have to use semiconductor grade, which is absurdly over-specified for the purpose and priced to match.
For a long time the PV companies could go to refiners’ back doors like hobos and buy at a discount the seconds, the ingots rated substandard for the real semiconductor customers, but now the demand has shot up so PV has to pay full whack. This is by far the biggest constraint on the future of PV. Making the panels is straightforward : the industry cry – just Google it – is “silicon feedstock”.
People are of course working on finding a specific route to medium-grade silicon at $20 or so a kilo. Whoever gets there first will make a fortune and save the planet like Superman, so it’s an attractive opportunity. The problem was entirely predictable given the relative growth rates. So why didn’t it attract much effort until recently?
I think there has been an institutional market failure. The challenge is out of the technological reach of the bulk silicon people; and the semiconductor refiners have I think been fixated on keeping Intel and company happy, customers who must be insensitive to price and fanatically demanding on quality. A $100 processor might be built on a gram or less of silicon wafer, or 5 cents’ worth – hardly worth worrying about compared to cutting rejection rates for the circuits.
You would think that this geopolitically strategic problem would attract oodles of public research – a money cake like Alice’s with EAT ME “beautifully marked in currants”. Not so.
The EU put €42m into PV research in the €17.5 bn 6th Framework Programme (2002-2006), with one €2.6m programme on silicon feedstock (search for FOXY): the 7th Framework programme hasn’t been approved quite yet, but funding for renewable energy will go up.
The USA, heart of the world semiconductor industry, spends even less. The current DoE programme for PV technology offers $12.5 million over 2-3 years . Searching the DoE site with the keywords “silicon feedstock”, I found precisely one grant awarded – for a princely $99,928.
This is the order of money the US government hands out as charity to cranks. The Pentagon used to support serious gravitational physics with the blue-sky hope of finding antigravity, and apparently funded some antigravity devices – well into the crackpot zone. NASA spent $1.6M between 1996 and 2002 on a similarly starry-eyed “Breakthrough Propulsion Physics” programme, in hopes of a real warp drive. (more).
You tell me this planet’s falling apart? Beam me up, Scotty.