Feb 20, 2007

solar energy

With all the news about oil and global warming there has been a lot of talk about renewable energy sources as of late. Ethanol or plant-based energy is the one I read about the most. Wind power a little and solar probably the least. I've done a fair amount of research on some of these and there is only one main conclusion you can come to at the end of it. We will all be using solar in the not-so-distant future.

Right now it costs around $4 to buy a photovoltaic cell that produces 1 Watt of energy. It takes another $4 to install that Watt. This includes things like labor, the casing, wiring, etc. The break-even analysis for a $8/Watt install means electricity would need to cost 40 cents a KWh for you to be parity with the grid. Grid electricity actually costs around 10 cents. So solar energy costs 4 times what it takes to produce electricity with other cheaper means (primarily coal). So why am I so bullish on solar?

Take a look at this graph. It's a plot of the average cost to produce a photovoltaic cell that creates 1 Watt of energy over time (orange line). The blue line is the cumulative production of cell wattage mankind has produced. This is an awful chart. It should be in log form and it's preferable to plot the average cost against the cumulative production. Nevertheless it's fairly easy to see that the cost has come down and it has come down exponentially. Basically there is something akin to Moore's Law for solar manufacturing. The more you produce, the better you get at producing it, the cheaper it costs. No other energy source is on this kind of price drop curve. End of story.
Right now there are a ton of companies starting up that produce solar cells. It seems like I get a new IPO prospectus on my desk once a week announcing another company is going public. Why so many? Well making solar cells is about as hard as making boxes. The equipment is standard off-the-shelf stuff and you don't need to raise a lot of capital to buy the equipment and start making it. But that's only half the equation. There must be some demand too. Why is there so much demand if it's not economic.

For the most part it is due to Germany and Japan. Both countries have set up extremely lucrative subsidies for people to buy solar cells. They have different approaches with Japan subsidizing loans and Germany subsidizing feed-in electricity rates (the rate the utility company pays you if you generate electricity). The U.S. is third but really we're talking about California which has also set up a subsidy program. Without those there would be no major increase in demand.

There's only one problem with all of this. At least for now. And that is silicon wafers. Kind of surprising since we've been making it for so long as an input to semiconductor manufacturing. The problem is no one expected these subsidy programs in Germany and Japan to have such an effect on demand. And a solar cell is a hell of a lot larger than an Intel CPU. 10 years ago silicon crystal demand from the solar industry was a tiny fraction of the total demand. Today both the solar and chip industry use about the same amount of crystal silicon.

So why don't the companies that make silicon crystal just expand? Well they could. And easily. The problem really comes from their input - polysilicon. This is the extremely pure precursor to making silicon wafers. And it turns out this is hard to make. Really hard.

To get an idea of why, consider that the polysilicon has to be anywhere from 9-9's to 11-9's pure to make a viable wafer. "9-9's" means 99.9999999% pure. That sounds difficult but not that difficult. We make lots of things extremely pure. It's not that hard. You just put it into a distillation column (same concept as making moonshine or purifying crude oil into gas or purifying uranium) and repeat, repeat, repeat by making tons of these distillation columns in a series.

The problem comes from one simple fact - polysilicon is a solid. Distillation columns only work when the purifying compound is a gas. To make polysilicon into a gas you mix it with hydrochloric acid and heat it up to very high temperatures. So not only is it hellishly hot, it is also highly corrosive. To make a distillation column that can withstand that is something of a secret. Only 7 companies have the real know-how to make a distillation column to do this. And they've been doing this for 30 years. Oh yea and they haven't had to build a plant in essentially the same amount of time. Until now.

Oh yea. And to make things just a little worse the one impurity you really don't want in your polysilicon is metal. Guess what distillation columns are made out of. So the only solution is to coat them with something. If a little of that something chips off and exposes a screw made of metal, your batch is toast. Now consider you are standing at the end of your plant and you realize your polysilicon batch is contaminated. You assume correctly that the coating came off somewhere. The question now is where? You may have hundreds of distillation columns in a series making this stuff. How the hell do you find a small speck of coating that came off in that? I have no clue.

On top of that there is a big lag between knowing you want to build a polysilicon plant and actually getting the thing running at speed. Something like 2 years.

So we have a polysilicon shortage. A bad one. Take a look at the only public company that makes polysilicon - WFR (click on the 5 year chart). This is a company that's been on the verge of bankrupcty many times. Most of the larger solar cell companies have paid these 7 polysilicon manufacturers to build new plants in exchange for 10 year polysilicon contracts. 10 years!! That's a basically risk-free proposition for them. That's unheard of. It gives you some idea of how much demand these guys are expecting and how sure they are that polysilicon shortages aren't going away anytime soon. Some day the supply shortage will alleviate but I don't think it will happen for a very very long time.

One reason is because when solar energy hits parity with the grid costs then it will make sense for anyone with a rooftop to buy a solar cell. There are a lot of rooftops out there. We'll go from only green companies and green individuals wanting solar to everyone wanting solar.

There are some other technologies. Other crystalline based compounds besides silicon. Thin film technologies that aren't as efficient but much cheaper. But these have their tradeoffs. If the efficiency of thin film isn't high enough you may not have enough rooftop to make all the energy you want. And thin film may not be as rugged as silicon-based cells which we know last 30 years and are required to have that lifespan to make the economics work out. But no one knows. These issues could all get worked out.

So there are a few problems in the way. But at some point solar will hit parity with the grid. And at that point I think we'll have a free-for-all. We'll see.

1 comment:

Anonymous said...

Excellent article. I appreciate your depth of knowledge in this field. I guess you must be a specialist in semiconductor field. I would like to keep in touch with you. May I get your e-mail address? mine is apjdus@yahoo.com