Last week, based on an IEEE photovoltaics conference that had just taken place in San Diego, I reported on some positive developments: a kind of Moore's law allegedly at work, a better outlook for first-generation silicon cells, and a growing expectation that solar electricity may be commercially competitive in some parts of the world by 2015. It appears that I misstated and overstated the case for a photovoltaic Moore's law, having--I'm most embarrassed to say--misread one of my own previous blog entries.
I said last week that PV costs per watt had dropped from about $7 in 2004 to $4-5 in 2007, but "Hal" pointed out in a comment posted May 22 that my previous blog in fact cited 2007 PV costs of $7.6 or $6.2 (depending on whose numbers you believe). That would be tantamount to a cost reduction of 15 percent at most, and perhaps none at all--not the 40 percent reduction I thought I had discerned, consistent with the "photovoltaic Moore's law" that postulates a 20 percent reduction with every doubling of cumulative production capacity.
In recent years, total capacity has been doubling about every 18 months, but longer term the rate has been more like 30 months, according to John Benner of the National Renewable Energy Laboratory, who takes some credit for originating the PV Moore's law postulate. Benner says the three years from 2004 to 2007 were anomalous because of a run-up in silicon prices and, perhaps too, the deteriorating value of the dollar.
Even so, relative to wind, photovoltaic costs do indeed appear to have improved rather dramatically between 2004 and 2007, from 7:1 to 4:1 or 5:1, close to 40 percent. That's significant, as distributed PV only needs to be about half as costly as central wind to be competitive. But PV's better standing owes as much to an escalation in wind's costs--connected with booming global demand for turbines and the general run-up in power plant costs, most likely--as to any cost reductions of its own.
The solar-wind comparison is a reminder that the photovoltaic Moore's law is subject to market forces in a way that the real Moore's law is not, as comments on my May 17 blog emphasized. Even the semiconductor Moore's law is of course not a law in the normal sense; it's a prediction that happens to have been born out, so far, by reality. But it at least is stated strictly in terms of physical parameters (how many transistors per unit area of chip). The PV Moore law is stated in terms of costs, making it subject to extraneous market forces.
"It used to be an axiom that solar power grew steadily cheaper as time passed," The Economist magazine noted last April. "After all, it had done so reliably for the previous 40 years." But in the last years, Germany's extravagant solar subsidies have driven up global prices for silicon, from $25 per kilogram in 2003 to around $400/kg now. "We didn't have a silicon shortage, we had an over-stimulated market," comments Benner.
Even if the consternation the higher silicon prices have aroused among solar boosters is somewhat misplaced--after all, just about every other energy cost is sky-rocketing too--the situation has prompted arguments that solar subsidies should be eliminated in favor of R&D support.
Is there an optimal rate of PV growth? If we try to give the industry too much of a kick can that backfire, disrupting the learning curve that's supposed to yield "grid parity" by middle of the next decade? Should research and production incentives be targeted or technology-neutral, and what is the right mix? Who provides the better model, Germany or Japan? Those were just the questions under discussion at the San Diego PV Accelerator Forum, and which will continue to be discussed by all who have the future of solar electricity and renewable energy at heart.
Meanwhile, I apologize to my readers for having overly accentuated the positive. I hope I read your writing more meticulously than I sometimes reread my own.