Solar Grade: A Silicon Revolution

Authors will do just about anything to catch your attention here. We're talking money. I suggest you get yourself a nice soda can of something and read this. But before you trash it, ever wondered what your empty can was worth like a 160 years ago? And what on earth has that got to do with solar energy stocks you might ask?

For those of you not too familiar with the technicals of the silicon solar Photovoltaic industry, I would like to bring to your attention several developments that in my view represent a paradigm change, will reduce the enormous volatility, present a view of the sector that could alter what I consider to be a false perception in the retail investment community about solar energy stocks in general and the silicon PV sector in particular, and hopefully contribute to some wealth creation -cq. to prevent losses- in a series of articles.

Rising energy costs and stretched power grids as well as a desire for energy independence and environmental/health concerns have sparked a surge in the use of solar panels to make electricity. Solar power (including hot water systems) is doubling now every two years. Doubling every two years means multiplying by 1000 in 20 years. That is the nature of exponential growth. Solar PV technology is an information technology and therefore subject to Moore's law which measures doublings in price performance. Solar PV technology has such doublings every six years. So why are we solar investors not all rich yet, boating around the Bikini Islands?

Because, for various reasons, there is a shortage of silicon refining capacity (but, unlike non-renewables: not of the raw materials that are used in the refining process) causing a statistical deviation of the main laws. Over 90% of solar panels in the market today use refined, purified silicon as raw material. In the 1970's, also a time of energy supply uncertainties, the solar panel market was created using scrap metal silicon from the growing semiconductor industry which uses ultra pure 99.999999% (8 9's or 8N) or even more pure 9N silicon. The scrap metal was pure enough: about 99.9999% (6N), less pure varieties were blended to maintain quality standards. Most solar cell production lines today need 6N: Solar Grade or SoG.

Then there is Metallurgical Grade [MG] used by the steel and metals industry as an alloying material. MG (98-99% pure) is made from relatively crude materials (silica or sand, coal or cokes) by heating in a furnace. The better MG is then the starting material of all refineries to produce ultra-pure 8N silicon in a gasification reaction called the Siemens process. Problem is these plants are very capital intensive, take a very long time to build and are very expensive to run. Then the end product is really too pure. Of course it can be blended, but all this remelting, heating and cooling causes futher inefficiencies, away from an optimized silica-to-solar-cells-process. All this caused governments around the world to subsidize the sector.

Spot prices for ultra pure silicon have reached $500/kg this summer. Contract prices (which are lower) are like discounted futures on the underlying commodity. So it usually leads the trend. Right now contract prices are decreasing slowly as more supply finds its way to the marketplace, while the spot has peaked. Of course future trends can also reverse themselves and maybe contract prices for 2010 and 2011 will start going up again, while the spot price is bottoming. The latter would be double bad news for the existing silicon solar PV sector as high spot has kept newcomers from entering the marketplace while even higher contract prices is a tough sell and could hurt margins.

High spot prices have also caused a large increase in R&D spending for alternatives to the Siemens process. REC of Norway [OSL:REC] is playing with a technology called fluidized bed reactors, which offer a power advantage, altough the process is not quite well understood, somewhat unstable and uses more expensive catalysts. Globe Specialty Metals [LSE:GLBM], Advanced Metallurgical Group [AMS:AMG] through subsidiary [TSX:TIM] have all reported succes upgrading metallurgical grade using pyrometallurgical processes, which are not as capital intensive. These stocks have sky-rocketed as they are building plants and/or ramping up production, even as it is still somewhat unclear what grade, when and how much. Dow Chemical (DOW) is also using such methods. Dealing with molten silicon remains difficult, especially the removal of Boron and Phosphorus, which are very detrimental to the functioning of solar cells in concentrations higher than about one in a million (6N)

But as every good cook knows, it is best to sieve the flour before baking the cake. Silicon can also be found as silica dissolved in water, in a cheap material called waterglass. It has been known since the 1990's that you can easily filter out many impurities (excluding boron, phosphorus) with reagents. But the real breakthrough came this century as Rohm and Haas (ROH) had developed some chelating (ion exchange) resins with usefull functional groups for the mining industry. But it took the genius of Steve Amendola to put it all together in a world-wide patent and start Reaction Science inc. www.rsi-silicon.com Production costs are three times less compared to a typical Siemens plant. Additionally the capital costs for building a chemical silicon purification plant is 1/10 the cost of a comparable Siemens process plant with a much shorter completion time. The fast plant fabrication time and easy scalability is another factor contributing to the low price of refining, along with the high yield. The ROI time on such plants is projected to be less than two years. The RSI proprietary process has been demonstrated on a small scale and vetted by leading Princeton and M.I.T. scientists. The process and cost estimates in the plan were validated by Black and Veach, a leading engineering firm. www.bv.com RSI already has orders exceeding .5 $ billion, customers include BP-solar. As a result the price of silicon PV solar products will be able to drop significantly. Plants such as these in Alabama, Europe and China will obviously benefit the entire silicon solar PV sector. The RSI method will also be used for forming mono-crystalline silicon.

Based on similar disruptive metals processes in the past, I am confident with a probability model, predicting a 95% reduction in the cost of refined (6N) SoG-silicon in 15 years from today's spot price.

Using Suntech (STP) as an example: silicon costs per watt were between $2 and $2.20, nonsilicon costs are about $70 cents or $80 cents/watt. In the next five years RSI will reduce silicon costs to $25 cents/watt and Suntech's non-silicon costs should fall to between $50-60 cents. Futher developments such as a household switch to DC, reducing installation costs (a union stronghold) and innovations/mergers could easily reduce costs even futher e.g. by energy savings.

Of course these positive developments are very bullish for the silicon solar PV stocks.

Some further thoughts:

Avoid ETFs in solar energy stocks such as TAN or KWT as they include some questionable equities with relatively very high market capatilizations. Solar energy investors should now choose sides after having weighted the facts, tracking the SPI and an ETF doesn't do that for you. I will discuss First Solar Inc. (FSLR) and other thin film stocks in another post.

In general it is cheaper to protect yourself from bear markets by buying puts in the general economy, rather then the continuously volatile silicon solar PV stocks, altough increased silicon price visibility should reduce volatility.

Well by now you should have finished your soda pop, the empty can was worth about $830 back then. A bar of aluminum was exhibited as a precious metal alongside the Crown Jewels at the Paris Exhibition of 1855.

Comments

[lucidlife3]

Excellent article, Mr. Van Wingerden. If only a third of all articles everywhere could be so substantial, and full of relevant information. Great work, lep

[JohnM]

The cell cost is not critical measure for solar deployment, especially those for utility plant scale. It is the system cost. Compare CPV solutions, i.e. Emcore, to those based on silicon, thin film, etc. And NO OTHER technology vendor offers a 20 year warranty.

[TellAVision]

Indeed, a lucid article if not exceptional blog post, but mostly old news to me. And yes, Steve Amendola is a genius. How do you get in touch with seeking alpha authors privately?

[vpratt]

This has been helpful. I look forward to more. If possible, I would like to hear more about your reservations about ETFs. Does your reasoning apply to regular mutual funds in the solar sector? Thanks.

[sopot]

You should have mentioned ARISE TECHNOLOGIE'S (APV - TSX) proprietary technology for producing 7N silicon around 40% less expensive than Siemens' !!! Arise got it's technology validated ( got already 2 contracts for 400 MT ) and will start production in the beginning of the 2009 from its pilot plant in Waterloo, Ontario.

[mistermaumau]

The way I see it this is bad news for FSLR and some as it will destroy the advantage they currently have over classical solars, do you agree?

[Road Runner]

Good information. Thanks.

[Jim Brown]

Minor point - 2 to the 10th is 1024, so doubling for 10 years would be a factor of 1000; and 2 to 20th is 1 billion odd.
Have you seen guntherportfolio.blogs.../ where he shows videos from InterSolar presentations last week in San Fran where they can keep up with Silicon (and its relatives) demand, which also keeps growing. They say the supply pinch is mostly inventory and logistics issues. And they can increase supply much more, as needed.

[Van Gastel]

Great article. Eagerly look forward to your next article.

Like mistermaumau I see this rather as validating the silicon solar panel technology, not really a traditional paradigm shift as it is not a new technology replacing another. Makes the outlook for FSLR look very bleak, so why do they still get these high ratings.

Also my comment on the oil prices. The "falling" oil price (very relative) is seen as tempering demand for panels, but all I really see is shipping costs for panels from China coming down...

Regards

[Jack Yetiv]

In additional to decreased cost of silicon, I also expect that watts/gram of poly will probably double in 5 years from todays.

This will be due to two key things: (1) increasing efficiency from about 16% today to mid-20's in 5 years (SPWR should be rolling out 23% panels next year), and (2) thinner wafers.

So even if poly prices did not change, cost of poly per watt will decrease substantially over the next few years. Factor in a drop in poly costs, and overall panel prices are well on their way to costing half as much as they do now within a few years.

Jack

[User 226214]

"then there is Metallurgical Grade [MG] used by the steel and metals industry as an alloying material. MG (98-99% pure) is made from relatively crude materials"

So far, I know only CSIQ producing Solar PV using UMG material which is much cheaper than the regular silicon.

[skwestorange]

Good info. More such in depth discussion of technologies by those knowlegeable in the field will be helpful

[lcg662]

Alas,

Has anyone wondered where all that absobed heat goes from the projected hectares of solar farms?

For every action there is an equal and opposite reaction; how many windmills does it take to tilt the earths axis?

Uhm... yes Emperor your new cloak looks very PC!

[aquaculture]

Thx all for kind comments.

JohnM: Concentrator PV will probably be cost effective in larger applications. My purpose is to compare the various solutions as they come up in a series of articles, as data come out.

Vpratt: The ETF issue will come up again in an analysis of First Solar inc.

Tellavision: Yes the news is old. This post had been ready long before the Rohm and Haas thing. Usually when I do a research and something like this comes up I have to act fast, as I have noticed I'm not far ahead of Mr. Market. Now that I'm set up with an authors page, I will publish more immediately, then we'll see what happens. But Rohm and Haas might have been the last best opportunity to profit from the silicon supply crunch, and I think now is a good time to get into the silicon PV solars.

Sopot: thx for pointing out Arise's efforts: I will look into it. Of course, tough small by comparison this only adds to my thesis.

Jim: If you read my post carefully it states a doubling every TWO years. This means ten doublings in 20 years, roughly a thousand times today's market share. By implication, if the street values these companies correctly, a mere $1000 investment would be worth one million $ in 20 years.

Jack: Silicon PV has great efficiencies especially when compared to thin film. The maximum is around 29 %, still a great price performance. The wafer thickness is limited by solar cell size. The market average is about 9 grams/watt. This leaves room to downsize. Then Evergreen (ESLR) has an innovative technology of 5.5 grams/watt. They expect 4.8 by 2009 and 2.5 grams/watt by 2012.
Of course all off this adds to my main thesis.
But which individual stocks to pick depends (increasingly) on future SoG-silicon prices. RSI will start producing at the 5000 mt Pennsylvania plant @ start of Q3 2008. This production is sold out so won't affect the market. On the other hand a 24000 mt plant such as this could be build in China in a few months.

User 226214: The term UMG is somewhat misleading. The Siemens process upgrades Metallurgical Grade silicon by gasification. Timminco (CSIQ's supplier) upgrades MG by remelting the 'dirty' silicon in a furnace and cleaning it up, with less capital upfront. Timminco can deliver 4N, 5N and 6N. It is CSIQ's luck to have been involved from the start. The amounts are still relatively small.
But by 2015 half of all silicon used in solar panels will probably be produced by such methods.

[aquaculture]

Aquaculture=Edgar
thx again for comments.

[alb1309]

Excellent article, Mr. Van Wingerden, will LDk's plant be in big trouble? How can they compete with RSI's plants?

[asd]

Looks like another scam company popping up, I wonder why they don't mention cell efficiency at all?

[sloemoe]

Moore's Law, as I understand it, does not apply to costs.It applies to tech capacity or performance of technology??

[nmcotwpv]

I also think this is an excellent article. I have thought for some time that most people underestimate silicon PV's ability to remain competitive.

On the other hand, I agree with User 232175 that Moore's Law (or some equivalent) does not apply to PV. Moore's Law states that the number of transistors in an integrated circuit doubles every two years. The cost of an integrated circuit has remained more or less constraint for the past 40+ years while its capabilities have increased in line with the number of transistors it contains. (As far as I can tell, Moore's Law is just a logistic function (S-curve) with nearly unlimited growth potential.) For there to be a solar equivalent to Moore's Law, there would need to be a nearly an unlimited potential to increase its efficiency.

[aquaculture]

Again thx for intelligent comments.

alb1309: You noticed I skipped the issue of LDK. On the one hand existing silicon PV shareholders will benefit from LDK's additional supply. On the other hand LDK shareholders carry the burden of an electr. grade silicon production facility with all the risks of possible delays, cost overruns. Then there are risks of production disruptions, fires and the like a-la WFR. I am not invested in LDK. An investment in LDK has become complex. I am considering an option strangle on LDK, but haven't done the research yet.

asd: which scam? what companies?

sloemoe, nmcotwpv: Fundamentally it is now generally recognized that information could be thought of as interchangeable with energy and we can measure its contents. But fundamentals aside, i did not state that solar PV= Moore's law, but is subject to it. Moore's law itself is indeed about squeezing twice as many transistors onto an Integrated Circuit every two years. This is an inverse to the solar PV surface area which doubles in same period. Beside the fact both technologies operate DC and presently use (mainly) silicon, given that the electrons would consequently have less distance to travel in order to do usefull work, there is exponential growth in the price performance of compution as well as the solar PV sector. That is how they are related and how these laws can be derived from each other.
Another way of looking at it is the fact that solar panels reduce the entropy of a system and we can measure that.
That there will be juice from the grid out of your sockets tomorrow is a hypothesis and this means: you don't know. The issue here is going to be finding secure means of making electronic information available. This is a fundamental economic concern. Google does not see the solar PV sector as separate from information technology. In fact they are integrating both technologies and there is enormous potential added value in doing so.

The paradigm shift described here is a production method of solar grade silicon allowing continued exponential growth of the solar sector. Each time a paradigm has run out of steam another has picked up speed.
My mean point is indeed silicon PV's ability to remain competitive.

[aquaculture]

aquaculture=Edgar

[aquaculture]

typo: compution=computation

[alb1309]

Mr. Van Wingerden: Thank you very much. When will you publish your research on thin films? Are you going to talk about fslr?

[buystocks]

all the solar stock will pick up. TSL with the best value now.

[aquaculture]

diatribune.com/breakth...

[TellAVision]

Edgar, are you SolveClimate, author of article at diatribune? didn't see an = sign... ~:-)
In any case, the McCain Dipstick is hilarious and pathetic at the same time.
Thanks also for your replies to commenters by name and my new shortened blogger moniker I've adopterd.

[trumanburbank]

Moore's law as applied to PV market is not a particularly useful or exact fit. See the article (Is PV Moore's Law Really On Track?) here:

blogs.spectrum.ieee.or...

[aquaculture]

Tell:
Fully endorsed, dipstick and all.

truman:
Without getting into tech talk here Moore's law has performed very well over the past 40 years ( and made stock holders a great fortune) because it is based on a fundamental law of nature: the learning curve.
It expresses a constant % improvement in some performance metric each time the cumulative number of trials or practice attempts is doubled.
This is a no brainer: we learn by doing.
Economic drivers explain why Moore's law exists.
Learning curve theory explains how.

'The cost per unit decreases by a fixed percentage every time the total cumulative output volume doubles when measured in CONSTANT currency'

Taking measurements of more than 35 years the learning rate for solar PV has been an exceptionally stable number: 20 %, right in line with...computers.

[brenkov]

RSI sounds like a winner. How would I get in before they go public?