Ask and you shall receive... Paul Maycock sent me his data free of charge and we even talked at length on the phone.User:Mrshaba/Experiments#PV Graph I wrote to Perlin today and I think he'll provide me with PV info for the early 70s or at least lead me in the right direction and clear up some inconsistencies. You seem new to WP but knowledgeable at the same time and you might be able to run with me judging from your posts. You might be able to outrun me... who knows... check out the info though. Mrshaba (talk) 17:44, 10 December 2007 (UTC)Reply

Yes, I'm new to WP -- I didn't even realize I had a user talk page until I noticed a little blue "talk" next to my ID -- but I'm "in the business," as they say. I'm getting more comfortable around here, so I'm hoping to do less talking and more contributing before too long.--Squirmymcphee (talk) 22:29, 12 December 2007 (UTC)Reply

I replied to your comments on the PV discussion page. Thanks for correcting my terminology. Maycock had to correct me on prices vs costs. I'd appreciate it if you can take a look at the price data. I'm doing my own cross check to make sure I converted current and constant dollars correctly but you might be able to eyeball the numbers. Mrshaba (talk) 20:28, 12 December 2007 (UTC)Reply

The influence of semiconductors on progress in PV

edit

Hi again Squirmy... I know I've been knocking on your door a lot lately so I hope you don't tire of my questions.

I think you might be working on a rewrite of the Solar Cell page so I figure you might be able to answer a question. I'm working on a rewrite of the photovoltaics section on the Solar energy page. I still need some more price numbers for the early 70s but I'll find them over the next few weeks. Thanks for the leads by the way. Most of the time I see the development of PV explained in stages... The space race stage, then the oil crises, the cheap oil years, the rebirth in the 90s and the modern explosion era. I've got these developments sketched out. One thing I'd like to give more weight to is the influence that semiconductor developments have had on the PV industry. To tell you the truth I'm mostly working off a gut feeling here but I don't think the influence of parallel development has been given proper weight. What do you think were the primary drivers of terrestrial PV in the early 70s just before the oil crises. I figure it partly had to do with NASA funds drying up and partly with semiconductor progress. Perlin gives most of the credit to Dr. Berman but I think at least part of Dr. Berman's success is attributable to timing. It seems to me progress in the semiconductor area has had an invisible positive influence on PV. How strong do you think this influence is?Mrshaba (talk) 18:40, 13 December 2007 (UTC)Reply

I wouldn't say I'm working on a rewrite of the solar cell page, but I've noticed some errors and inconsistencies that I'm slowly trying to fix (for the most part I think it's a pretty decent page). I am seriously considering a top-to-bottom rewrite of World energy resources and consumption, though.
At any rate, while Perlin's book is interesting I think he sacrifices a lot of the big picture for the sake of storytelling. Have you read Solar Revolution by Travis Bradford? His coverage of the big picture is much better than Perlin's, though it probably goes a little farther in the opposite direction (too little detail) for what you want. He attributes a lot of the initial interest in PV to the first oil crisis in the early '70s -- up to then, to my knowledge, terrestrial PV was largely experimental. If you have access to the Proceedings of the IEEE Photovoltaic Specialists Conference from the late '60s and early '70s, you will find almost no mention of terrestrial PV until around the time of the first oil crisis (you'll probably have a hard time tracking those down, but honestly they don't really contain much of the information I think you're looking for). There's also no question that PV was embraced by the early environmental movement, though since PV was so expensive in those days their support was more a matter of policy than economics. You might also look for Harnessing Solar Power: The Photovoltaics Challenge by Ken Zweibel, which I first saw at a public library about 15 years ago. It's outdated, but as I recall it provides some historical context. For that matter, your best bet at a contemporary, 1970s viewpoint might be to dig up some of the many books on energy that were published in the late '70s.
As for the influence of semiconductor developments, I'm not aware of a good resource that summarizes it (though one may certainly exist). I think virtually everybody would agree, though, that the single largest influence is that it enabled silicon feedstock to be purified and sold cheaply. Perlin's book has a quote from Daryl Chapin (if I recally correctly) where he estimates the cost of a solar cell in the late '50s and says its high cost is due primarily to the high cost of silicon. Adjust Chapin's figure for inflation and compare it to the current price of silicon feedstock, which is roughly $75/kg (and can be sold profitably at $30/kg). I think you'll understand my point. Beyond that, the PV industry pretty much relied upon the methods and equipment developed by the microelectronics industry for quite a long time -- until you have the capital to develop your own equipment, there's not really much else you can do. That was good from the standpoint of keeping costs down, but it was bad in the sense that a lot of laboratory developments could not be transferred to the production line. That's changing a bit now -- the PV industry is flush with cash and has outgrown (quite literally) much of the microelectronics equipment it has grown up on. You now see companies like Centrotherm, Applied Materials, RENA, GT Solar, and Spire, to name a few, catering to the specific equipment needs of the PV industry, and that has enabled the industry to be more creative with its processes. So in a nutshell, I think the influence of the semiconductor industry has historically been very strong -- and I doubt you'll find many people in the PV industry who would disagree with that statement -- but I think it is beginning to diverge a bit from that path. I think the divergence will become greater with time, and I think the PV industry might eventually influence the semiconductor industry in positive ways (which it has already done via the wire saw for wafer slicing), but I don't think the influence will ever entirely disappear.--Squirmymcphee (talk) 20:00, 13 December 2007 (UTC)Reply
Thanks for your response. I will read it several more times. I think we are like minded in how the development went down moreorless. Recently I purchased 21 bound copies of Solar Energy from the Ames Research Center. Unfortunately, the volumes don't overlap with the late 60s and early 70's but I do have the 65/66 volume. I found an interesting paper today: "Design of Solar Cells for Terrestrial Use" by Paul A. Berman (Senior Engineer, Heliotek, 1966). Here is a quote: "The cost per watt of present-day space power systems is about $1000 per watt, with the cost of the solar cells alone accounting for about 20 percent of the total. These systems, however, are subject to critical weight, area, and deployment considerations, which would not be necessary for terrestrial systems." The paper goes on to say if we do this, this and this we can get the cost down to $36/watt. The this, this and that had more to do with tailoring PV to terrestrial needs rather than space needs. The references listed provide a good trail to follow so I'll have plenty to keep me busy. I think this paper and the others it leads me to will make a good case for what the developers were thinking. I'm sure I'll find some references which mention semiconductor developments trickling down. If you happen to find anything shoot me a note.Mrshaba (talk) 00:33, 14 December 2007 (UTC)Reply
I promise these are my last questions for the week. Are you an ISES member by chance? I've recently become a member but have had no luck finding any kind of message board for members. I have basic questions I want to ask. For example: I have this paper in my hand but I would like to know what other papers since published reference it. Sort of a reverse reference list. Do you know of a straight forward way to search for forward references? Mrshaba (talk) 05:16, 14 December 2007 (UTC)Reply
I'm not a member of ISES. I used to be, but it was long ago enough that they barely had a web presence, let alone a message board. I am a member of ASES, which has done a lot of work on it's web site over the past year, but I haven't really explored it well enough to know if it includes message boards. As for finding papers that cite a particular paper, the best resource I've found for that is Google Scholar. It's probably incomplete, so I also tend to Google on the exact title of the paper (enclosed in double quotes) when doing such a search. Some journals do pretty rigorous tracking of who cites their articles, so you might also check the journal's web page if the paper is a journal article. There might also be some central clearinghouse that university libraries subscribe to -- I don't know for sure.--Squirmymcphee (talk) 16:12, 14 December 2007 (UTC)Reply

Learning Curves

edit

What puzzles me about Berman's info is that his Solar Power Corporation was the first dedicated terrestrial PV manufacturing company. His information doesn't represent just any data point - it's arguably the starting data point. I cannot disregard this data without an explanation. The data really bugs me.

Going back to 1967, Paul Berman (no relation to Elliot Berman), the Senior Engineer at Heliotek wrote an article for the Solar Energy journal entitled: Design of Solar Cells for Terrestrial Use. In six pages he outlined simple ways to bring down the cost per watt of PV. Several others had written similarly on this topic but SPC was the first to manufacture product according to these principles. The end result was that SPC economized PV by de-optimizing modules. What I find strange is that the price they achieved was 4 years ahead of the rest of the industry according to the data Maycock gave me.

Additionally, SPC delivered product before OPEC was a household name. Sailboat owners, lighthouse operators and Telecom Australia were buying from SPC before any subsidies or energy crises existed. These were practical consumers and this is an important note. How could SPC pick the low hanging fruit that many others had pointed out only to have module prices escalate three fold in the following years while production went up several hundred fold? To me, manufacturing volume doesn't explain this time period. The last several years are also an anomaly but I agree with your reasoning for this departure from the curve.

Learning curves are helpful as gross tools but they've been presented as prediction tools. Statements like: If we increase production by a factor of X we can get the cost down to Y. This is not true. It appears several ranges of data cannot be accurately explained by the learning curve model alone. All and all, I think the learning curve model should be applied conditionally. I will check out the paper by Nemet and I appreciate this reference.

I want to improve my understanding of the driving influences of PV over its history. I don't think boiling down individual characteristics into a mathematical expression is particularly useful. It gives you a synopsis but the story is better told by chapter. The thing that started this little conversation was the question of how to present the data graphically. You suggested a log-log scale. I've done that and also broken things into smaller linear chunks. Both methods work fine. I'm not trying to be hard headed with you regarding learning curves because you clearly know your stuff. I work in the energy field but on the opposite end.

I'm broadly interested in the solar energy field in addition to just PV. I'm starting to understand that solar energy statistics are out of whack. We don't put meters on our windows but this does not discount the sunlight that comes through them. I now see that solar energy forms the baseline for all the other energy inputs to stand on. The completely radical idea that has sprung into my mind is that civilization uses more solar energy than any other sort. I've never seen this idea in print... But as with all other decent ideas I've had I'm sure there are many many people who have put forth this idea.

Yes, skipping off subsidies like a rock on water. Floating on subsidies is probably a better way to put it. Subsidies and eventually the price of grid electricity will continue to buoy PV prices and costs. That's my prediction anyway... This is going to continue to throw off the learning curves' "predictions". If you'd like to continue this conversation off-line here's my email. It will be good for two more months. lakc@pge.com Mrshaba (talk) 04:42, 2 February 2008 (UTC)Reply

I don't think you're being hard-headed at all -- debates like these are quite useful, and this discussion is quite similar to those I've had with others who are themselves in the PV industry. Plus, they force me to think about my own attitudes and assumptions instead of just lecturing others about them. :-) What drives the PV learning curve is a matter of great debate, and I think the only consensus on that matter is that the driver changes with time. To my knowledge, Nemet is the only one who has systematically tried to attribute learning rates to specific drivers (though the industry obviously tracks and discusses cost drivers outside of the context of the learning curve).
The Berman thing isn't as troubling to me as it evidently is to you. As now, I think a lot of the companies that entered the industry early on had their own ideas about how to reduce costs. Solarex, for example, likely had high production costs because they were dumping capital into developing the cast multicrystalline silicon process. I don't know that this was the case with Berman, but it is entirely possible that his focus was on immediate cost reduction while others took a longer view and invested in technologies they felt would better position them for the long term. I don't think this is so unusual for an industry in its infancy. I'm also fairly certain that some companies in those days were trying to minimize their expenditures by adapting processes for space cells to terrestrial cells, a strategy that didn't last very long but might have bumped average prices upward. Finally, Berman and his competitors were also not likely sharing cost information with one another, so his competitors might not have realized what the difference was. That said, one of the things I've never asked anybody about is precisely where these numbers come from, as I don't think there was any formal surveying of manufacturers in those days like there is now. I suspect Maycock based his numbers on information gathered in the course of administrating government research contracts, but I just don't know.
I would have to agree that one must be careful when using learning curves as prediction tools, and for exactly the reasons you cite. There's nothing fundamental about them, they are simply a convenient way to describe historical trends. That said, they have been applied over a wide range of industries and I am amazed by how infrequently major discontinuities occur. Plus, anybody making predictions about the future needs a basis on which to build the prediction and, for better or for worse, that generally means historical data. Forecasts based upon highly stable historical trends tend to be more believable than those based on pure speculation, no matter how educated and well supported the speculation might be.
As for using it to detect cost drivers, I think the learning curve is only useful for that if you approach it as a statistical tool. Unless you do something like Nemet did, using a statistically inspired approach to correlate it with industry changes and trends over time, I don't think it is very useful in that regard. I can't tell you how many conversations I've had with other engineers where we sat around asking each other, "why does the curve look like this and what does it mean?" Of course, by now I'm sure you're well familiar with this feeling.
Going back to the log-log scale, the reason I prefer it is twofold. First, it is familiar to anybody who has studied learning curves since their inception (in fact, the mathematical relationship was originally derived from the log-log plot). In that sense, it's like a word that everybody understands -- it's okay to use a different word to represent the same idea, as long as you understand you might have to do a little extra work to make sure everybody understands your definition. Second, because the mathematical relationship is so intimately tied to the graphical representation I find the log-log plot easier to interpret. As long as everybody is on the same page and linear-linear plots support your claims there's certainly no problem with representing the data that way.
I agree that solar energy statistics are underreported, and I think it has to do with the definition of "solar energy consumed," the ease of data collection, and the perceived role of solar in our energy portfolio. The statistics published by EIA, for example, only count power plants 1 MW or greater in size. This isn't a big deal for most energy technologies, but you can imagine the potential for PV and its multitude of installations measured in kW to be seriously underreported going forward (I'm told that DOE is thinking about ways to correct this). If you define passive solar heating of homes as "solar energy consumed" you have a data collection nightmare, but because of the way we perceive our energy consumption most people don't even think about it. And of course, if you expand your definition far enough then all of the fossil fuels we use are really solar....
Finally, I'm not convinced that grid power prices will "float" PV prices. If the PV industry is competitive, nothing will float it -- the only reason prices are stuck right now is because of the combination of subsidies and lack of competition, in my opinion. Furthermore, worldwide retail grid electricity rates span a range at least as wide as about 3.5 cents/kWh to more than 35 cents/kWh (and even higher, if you count time-of-use metering), so "the price of grid electricity" has a rather ambiguous value when you're talking about a global industry. I think that for the foreseeable future, no matter how big the PV industry gets there will always be somebody paying less for grid electricity; expanding PV supply, then, would have to entail a price reduction unless there is unmet demand among customers paying higher grid rates (in which case demand likely outstrips supply and we're back to where we are now).--Squirmymcphee (talk) 22:07, 4 February 2008 (UTC)Reply

Thanks again for the lead on Nemet. Just the guy I was looking for. He points out that learning curve models should be used cautiously.

http://papers.ssrn.com/sol3/papers.cfm?abstract_id=946173

"This study examines the case of photovoltaics (PV)... The results indicate that learning from experience only weakly explains change in the most important cost-reducing factors— plant size, module efficiency, and the cost of silicon." Mrshaba (talk) 20:27, 2 February 2008 (UTC)Reply

Have you had seen the complete paper or just the abstract? Just as a point of clarification I'll say that the paper is framed as an investigation to determine how much of the learning curve can be described as "learning by doing" and how much progress can be attributed to concrete technological and economic changes. The learning curve itself originated with an airplane factory that essentially made the same product over and over again for years. The owner of the plant noticed that his costs came down even though he hadn't made any substantial changes or improvements, and he attributed the improvement to efficiency gains made through his employees' learning how to streamline their work over the years -- hence the "learning by doing." Nemet contends that learning by doing has not been nearly as significant as increases in plant size and module efficiency, and reductions in the cost of silicon. It's not a surprising conclusion when you think about it, but the paper is notable for the way in which Nemet analyzes the learning curve.--Squirmymcphee (talk) 22:07, 4 February 2008 (UTC)Reply
Thank you for the detailed response. I read through Nemet's white paper that I provided a link for above. It was about 27 pages plus an impressive list of references. It was impressive all around and I'll read more of Nemet's stuff when I get a chance. I would have liked more detail about silicon cost reductions in the 60s that set the stage for commercial PV in the 70s. I am still in search of this information because I figure silicon cost reductions helped push terrestrial PV to a threshold. Nemet pointed out that silicon costs are one of the primary drivers for PV price declines. I had already felt this and put it to Berman but he did not agree and responded that terrestrial PV happened because "we willed it". I can't argue with him because it's a wonderful response and it's possible silicon wasn't all that important during this time period but only acquired importance going forward. Interestingly, Bill Yerkes echoes the "we willed it" sentiment with his second lieutenant idea. Regardless of favorable or unfavorable conditions, things get done because people choose to attack a problem.
I asked Maycock and SU if their price numbers averaged together earth and space PV prices as this would have completely explained the data scatter. SU said in essence "we think so" and Maycock said he wasn't going to argue over one data point. Nevertheless, the averaging of earth and space PV prices is probably one of the main reasons for the scattering of prices and this follows your reasoning. So now I can start sleeping again. It makes sense that as space suppliers took more interest in the emerging earth market they didn't jump in with both feet as SPC did. You have to consider that they were perfectionists who had long lived off and by NASAs' lead and need. Maycock told me to ask Yerkes for more information so I'll do that. I'm also going to research Solarex and their early prices. These early prices concern me because they mark an important transition for PV into a real, relatively unsubsidized market. SPC had customers at a price of $20/Watt - summer cottages, lighthouses, navigation buoys and off-shore oil derricks. These customers shared remoteness in common and this made PV practical for them at $20/Watt. The point is, I want to be able to trust the price that made PV practical for these customers. That's why this data bothers me.
I'm thinking about the idea of PV islands. The basic idea is there are customers which can be grouped according to their unique characteristics. Germany might have near universal grid connectivity and be geographically connected to the rest of Europe but their feed in tariff makes them a sort of island. Spain and California fall into this subsidy island for the same basic reason. Another sort of island would be a location with favorable insolation coupled to high local grid prices. I think the PV industry will tend to concentrate on these islands and this is why PV prices will tend to float. Eventually PV will outgrow subsidies and compete with the grid in numerous areas. After this point the the high priced portion of the grid with favorable local insolation will continue to float PV prices. This condition should last for many years and it should drive outstanding growth.
As far as measuring solar energy use goes I agree it's very difficult for a number of reasons. It's too much for this communications but I have some ideas on how to do it. Some fashioned some borrowed. Cheers and goodnight. 63.196.195.15 (talk) 04:44, 6 February 2008 (UTC)Reply
Somehow I managed to miss the link you provided. The paper was published in a scientific journal at one point and I had no idea he had published it as a working paper, so having missed the link it wasn't clear to me what you had read. I agree that "we willed it" is a wonderful response, but the fact remains that silicon was insanely expensive in the '50s compared to now -- even at the current inflated prices -- and only came down with the growth of the microelectronics industry. By the early '70s it was much cheaper, but still not as low as today; factor in the thicker wafers and less efficient sawing methods of the day and it's not hard to imagine that wafers were quite expensive. That said, if Berman was buying scrap wafers, which was not unusual for the PV industry up to the early 2000s, then it's entirely possible that he got them well below cost. It's not as romantic a notion as his response implies, but it might be closer to the truth. Your "PV islands" idea is intriguing, though I'm not 100% convinced. I'll have to give it more thought.--Squirmymcphee (talk) 06:10, 6 February 2008 (UTC)Reply

By the way

edit

Here are some maps if you are curious. Google image search kwh/kw and you will find them.

http://www.eere.energy.gov/consumer/your_home/electricity/index.cfm/mytopic=10860 http://re.jrc.ec.europa.eu/pvgis/countries/europe/pv13_opty.it.png Mrshaba (talk) 16:37, 6 February 2008 (UTC)Reply

Is it you I've been discussing this with all along? The posts in that discussion are all signed with IP addresses....
Either way, thanks for the maps. I think they support what I've been saying quite nicely. Neither of them purports to be an insolation map -- the word "insolation" doesn't even appear on either page -- but they both purport to be maps for estimating annual kWh production given the size of the PV system in kWp. And as I just pointed out back on the original discussion page, if these were truly insolation maps then they would suggest that if you have zero kWp of PV then it must be dark outside.
I suppose this whole thing might seem like a pissing contest over a trivial matter, but technology is a difficult enough topic to discuss even when everybody is working from the same set of definitions. If a climatologist -- someone with expertise in global sunlight patterns -- with no knowledge of PV were to review an insolation map with units of kWh/kWp he wouldn't have a clue what it means. Furthermore, it would require a good deal of explanation to help him understand it. Is an insolation map really an insolation map if an expert in insolation can't read it?--Squirmymcphee (talk) 17:28, 6 February 2008 (UTC)Reply
I am not the annonymous one but I've butted heads with him. Track down his opinions on space colonies because they're hilarious. I happened accross the conversation is all and remembered having seen the maps. That's all... Mrshaba (talk) 21:43, 6 February 2008 (UTC)Reply
Ah, good to know -- I was worried that you had a Jekyll & Hyde personality! This is my second encounter with the person and I'm learning that I can't predict where his logic will carry the conversation. I'll be keeping my responses to him short and simple in the future, if I respond to him at all. I'll look for the space colonies....--Squirmymcphee (talk) 23:14, 6 February 2008 (UTC)Reply
I'll save you the trouble of looking. At some point in the future, bearing in mind that we know that we can live on the planet for a billion years (short of any catastrophe) and are not constrained to the planet, there will be more people living off the planet than on the planet. NASA says there is enough material in the asteroids to build space colonies big enough for a trillion people. Since we expect the population of the earth to level off at about 9 billion, we can expect that the tipping point will occur sometime in this millennium - in less than a thousand years. Stephen Hawking says we will only be safe from extinction when we have populated outer space to the point that if the earth has a catastrophe it can be repopulated from space. My views may be hilarious to the clueless, but they are really quite ordinary. 199.125.109.27 (talk) 05:52, 7 February 2008 (UTC)Reply
I have to admit, I don't really find that viewpoint so unusual. I don't find it particularly usual, mind you, but it certainly isn't novel to me (nor is it necessarily a view I share, but I'm not going to debate it, either...).--Squirmymcphee (talk) 19:25, 7 February 2008 (UTC)Reply

Can you guess?

edit

Today I read that Si spot market prices are $300-$400 per kilogram. Correct me if I'm wrong but IIRC each kilowatt requires 8-10 kilograms of Si so this means spot market Si leads to $2,500-$4,000 per kilowatt in silicon costs alone. In the same Greentech article they mention panels need to get down to a price of $2/watt before the market really takes off. I don't think this is necessarily a magic number but I'm wondering how low costs have gotten if you account for the exaggerating influence of Si. I would think the low cost producers would probably be down to $2-2.50/watt. Do you have a guess?

Do you happen to know roughly how much cheaper solar grade silicon will be compared to electronics grade? What do you think a reasonable long term price for Si is? $10-20/kg?

I often see the PV industry described as having low barriers for entry but I'm thinking the cost of silicon plus the impressive scale some producers are reaching presents a significant barrier for entry. When the costs of Si come down in a few years there could be several GW+ producers who have developed products with efficiences approaching 20%. Do you think the industry can still be described as having low barriers for entry? —Preceding unsigned comment added by 131.89.192.111 (talk) 17:39, 21 February 2008 (UTC)Reply

I missed the article citing spot market prices at $300-400/kg -- got a link? I don't doubt you, but I'm interested in the article.
A kilowatt of PV probably requires more like 10-11 kg of Si, though that number has been dropping rapidly since Si got expensive. Yes, that means making solar cells with spot-market Si is a very expensive proposition. It is mitigated by a couple of factors, though: (1) too little Si is available on the spot market to sustain significant production volumes and (2) major manufacturers generally obtain their Si via contracts with Si purifiers and/or wafer manufacturers, and contract prices are more like $60-80/kg right now. Spot market figures give you a pretty good clue about just how tight the supply is, but they won't give you a good feel for what companies are actually spending. I have it on pretty good authority that direct module manufacturing costs are currently in the $2.50/watt range, give or take, for most silicon PV manufacturers, and that the figure would be much closer to $2.00/watt if we still had the low Si prices of a few years ago. Manufacturers are usually pretty guarded about that sort of information, though, so it's hard to know how reliable those figures are.
The price that solar-grade silicon will eventually reach is anybody's guess. I've been told that electronics-grade silicon must sell for $30-35/kg minimum for it to be profitable to the purifier and that the methods companies like REC are currently exploring will only shave $5-10/kg off that figure. Then again, others are developing processes they think will bring the cost down as low as $15-20/kg. In the end, though, silicon is a commodity and supply & demand will set the actual selling price -- the price required for profitability is more of a tool for refiners to decide whether or not to build more production capacity. Long-term I think contract prices in the $25-35/kg range would not be unreasonable, but I think as PV companies grow they're going to begin to vertically integrate to the point that they buy sand do the refining themselves. A few of them are already laying the groundwork for this, and in that case the cost of silicon will depend on the refining method the company chooses and just how pure they want their silicon to be.
As for barriers to entry, I think it depends on what part of the industry you look at. The barriers seem to be rising across the board, but in some segments they're rising faster than in others. Wafer production and solar cell processing both require quite a bit of capital if you plan to do them on a large scale (and as you point out, the scale required to be significant is growing rapidly), but if you buy cells and ship them to a country with low labor costs to be assembled by hand into modules then all you need is a little working capital. Anybody with the right certifications can install PV systems for little more than the cost of the tools and the certifications.--Squirmymcphee (talk) 19:07, 21 February 2008 (UTC)Reply
Thank you for your insight. Your information is always very helpful.
I crossed my wires when I said a Greentech article quoted a spot price. Here's the actual link:
http://www.businessweek.com/investor/content/feb2008/pi20080220_903992.htm?chan=top+news_top+news+index_global+business

Mrshaba (talk) 00:12, 22 February 2008 (UTC)Reply

I've continued to think about the early era of PV and the information we batted around about Berman and Solar Power Corporation vs. Varadi/Lindymeyer and Solarex. I've collected several more early PV price data points from NASA documents and whatnot. In general the data shows Maycocks' numbers are high. I've pretty much put this down to a blending of data between terrestrial producers like Solarex, SPC and STI vs. space producers such as Spectorlab. I asked Varadi for some early price/cost and production numbers as well as other questions. While he never answered my main price/production question he did say in the early days Solarex considered Spectrolab their main competitor. This surprised me but if Varadi considered Spectrolab the primary competitor in 1973 it seems likely Spectrolab was included in the early price data of terrestrial PV. As I said before, Maycock and Strategies Unlimited could not confirm this theory. Another thought on this is that the inflated price data might have thrown off the early calculations on price reductions and the R&D expenditures required to drive price reductions. Oh well right... This is all old info but it helps to understand the early history of PV and PV's entry into the remote terrestrial market prior to the Oil Crisis. This is mostly overlooked but I think this information provides a lesson to today's remote market. The Kenyan market comes to mind and the work of Grameen Shakti in Bangladesh both show PV has nontraditional markets that have straightforward drivers relatively outside of subsidies and research efforts. Isolation always strengthens PV's position... thus my island idea...
But here's an idea leading to a question. It's an idea in process so forgive me some obvious errors. You mentioned that Berman produced the cheapest product possible by using space PV discards while Solarex took a longer view and attempted to develop a multicrystalline cast process. I think you are right but this led me to the idea of the sheltering affect PV has intermittently benefited from. In the case of SPC they used space PV discards but this process was not sustainable for a large market. This was the first shelter but PV continued to grow beyond this early shelter by swimming along the microelectronics industry which made possible a relatively cheap supply of silicon. At this point PV has again grown out of this other shelter. Figuratively speaking is PV still a child or now a middle adult? I would think the silicon situation shows that PV is starting to standon it's own. The obvious counter to this middle adult idea is all the subsidies PV receives. My question is... Can you see other significant areas outside of subsidies where the PV industry has or is currently benefiting from other strong sheltering influences?
One other question I've been meaning to ask. Is the industry as exciting as it was 30 years ago or more so? It seems the public perception is ho hum but I'm wondering about the industry perspective. For me the situation seems really wonderful. Huge growth and progress and a target that isn't all that far off. What do you think? Who has made projections that seem most likely to you? 66.122.72.201 (talk) 04:53, 22 February 2008 (UTC)Reply
Thanks for the link. Did I ever suggest you read "Experience Curves of Photovoltaic Technology" by Christopher Harmon (it's a white paper that I think you can Google pretty easily)? I thought I had, but if I did it apparently wasn't on this talk page. As I recall, his price data for the early days of PV is substantially lower than Maycock's (it's not really "his" data, but I can't remember where he got it).
In terms of islands, the off-grid market is definitely a different animal from the grid-connected market, and the off-grid market in the developing world is different still. Those sorts of islands I can definitely see -- in the context of our previous discussion, I was thinking only in terms of islands within the grid-connected market. Right now, the grid-connected market is the market driving the PV industry and all of the industry's goals seem to be defined in terms of grid prices for electricity. For years I've been telling anybody who will listen that the developing world represents an enormous opportunity for any PV company that can figure out how to tap into it. Right now I think PV companies see little reason to try, since they can sell everything they make without lifting a finger to do it (PV companies have the most bare-bones sales staffs I've ever seen and are still sold out for the foreseeable future). Once the market becomes more competitive, though, I hope that will begin to change. I think it's likely that localized, highly distributed power generation will be much cheaper than building centralized power plants and national grid systems, but to do that the locals themselves have to put up the money and maintain the systems. I would love to be involved in a project like that someday.
I don't know that Berman went for the cheapest product while Solarex took the long view -- that was speculation on my part. But PV has definitely benefited from what you call the "sheltering effect". Space discards, then microelectronics discards, and a massive influx of microelectronics industry knowledge all benefited the PV industry. Subsidies are, right now, the big one. If the subsidies were to end today, the industry would collapse -- there's no other shelter that could protect it. That's not to say that it wouldn't have grown at all over the past decade, but its growth would probably have been closer to its former 10-15%/year instead of its actual 40%/year. It's taking control of it own silicon supply, and you're seeing technologies like CdTe and CIGS that nobody had been able to commercialize for decades actually hitting the market. I think you could call the PV industry a young adult. When it starts to get serious about glass costs, that's your sign that it's maturing.
I can't say how exciting the industry was 30 years ago because I wasn't in it at that time, but I can say that it's an awful lot more exciting than it was 15, 10, or even just 5 years ago. I don't think energy is quite as important in the minds of the public as it was 30 years ago and I think the message they get from our press and our politicians is not as gloomy as it was then, and that may have to do with public perception. And some remember all the promises about alternative energy technologies from the '70s and might be a little wary of a repeat. At any rate, as long as current growth rates are sustained I think PV modules will hit the $1.00-1.50/watt range (retail) in 7-12 years. Whether that's enough for grid parity in the US will depend on how inverter costs drop and especially on how installation costs drop. When people focus solely on module costs I don't think they realize just how much it costs to install them, especially for small home systems. I think that will be taken care of, though, especially as modules are increasingly designed with ease of installation in mind and more installers pop up to create competition.--Squirmymcphee (talk) 16:07, 22 February 2008 (UTC)Reply
You formerly suggested the Harmon paper and I've read it. Harmon described his learning curve data as the most complete set yet collected but I didn't take away much from his conclusions and he's dropped off the map since the paper so I couldn't follow up with him. His data came from both Maycock and SU as well as some other sources. I've said before and I'll say again I think the story is better understood by partition rather than production. Despite the fact that the learning curve works I think it's an oversimplification.
I also understand that you were only speculating about Berman's approach but at this point I agree with your speculation. I'm happy to hear you agree the developing world represents a fantastic market for PV. I've only come to this conclusion recently but this is clearly an under appreciated possibility and one to track in the future. The phrase "good Karma" comes to mind when I think about PV and developing countries. The idea that a developing country can skip some of the non-renewable problems of industrialized nations through PV is great. Check out Geodestinies if you haven't read it already.
I agree with almost all of your projections and thoughts. The installation end is peculiar though. Here's a thought. Right now it's more economical to install PV from the get go on new housing with the subsidies geared around having professionals install PV (at least in California). In the not too distant future with a market absent restrictive subsidies the do-it-yourselfer will be able to install most of a system up to the hot-work and inspection. I figure manufacturers are obviously going to produce more easily installable panels over the coming years which will reduce installation costs on the integrated end as well as simplify things on the retrofit end. Some might argue this effect won't be significant but I think do-it-yourselfer installation might play into things. Hell, the Bangladeshis are putting up panels on bamboo which shows me installation can be approached from several angles. All I'm saying is that installation costs could be reduced through non-professional home owner labor and this could dampen some of the installation cost burden. In San Luis Obispo where I live there's a law that XX% of new homes in 2010 have to be installed with PV. I would rather see 100% of homes be installed PV ready and let dynamics take over from there. But who knows. The situation in just 2 years will shift so we'll just have to see. Always good talking to you Squirmymcfee. I think I've run out of questions and ideas for now but I'll get back to you. Mrshaba (talk) 05:19, 23 February 2008 (UTC)Reply

Cross check

edit

I've read several articles mentioning the slowdown of the PV market lately. I've taken a second and third look at Germany's Feed-in incentives vs. production costs and it seems to me the estimated "spread" is still handsome enough to drive more investment. It appears the Spanish market will genuinely cool-off but I don't see an overall slowdown. How do things look from your end? Mrshaba (talk) 19:48, 8 March 2008 (UTC)Reply

Any idea what Michael Rogol says about the prospect of a slowdown? So far it seems he has never been wrong about these sorts of things, even when his remarks contradict the vast majority of market observers. I've been extremely busy the last month or two and haven't had time to read up on what anybody is saying, but since demand has been so much greater than supply in recent years I don't think a demand slowdown would necessarily be a bad thing. Among other things, I think it would it would force competition in the industry and cause a rapid drop in retail PV module prices. A slowdown in demand might affect manufacturers' long-term plans, but I doubt it would keep anybody from selling their entire inventories for the next year or two.--Squirmymcphee (talk) 19:09, 11 March 2008 (UTC)Reply

There was an emotional article in the Washington Post yesterday concerning silicon processing related pollution.

http://www.washingtonpost.com/wp-dyn/content/article/2008/03/08/AR2008030802595.html

They use a fellow by the name of Ren Bingyan (School of Material Sciences at Hebei Industrial University) as a reference. It appears as though Bingyan might moonlight at the University but during the day he's a VP with Alcoa.

http://www.zoominfo.com/Search/PersonDetail.aspx?PersonID=960875021

Is there a connection between Alcoa and solar? The only connection I can think of is their common interest in cheap electricity for refining. Mrshaba (talk) 13:49, 10 March 2008 (UTC)Reply

Thanks for the WaPo link -- it's been going around. For the reasons mentioned in my last comment I haven't had time to read it yet, but based on the headline I get the impression that it is critical of a particular plant in China. That's hardly representative of the rest of the world, which is responsible for 90+% of the industry, or even the rest of China. I wouldn't necessarily discount the story's importance, though, because China's share of the market is rising rapidly and, especially in a "green" industry, reckless pollution by any player needs to be nipped in the bud. If the article is trying to demonize the entire industry as hypocritcal polluters, though, then it's sensationalist crap.--Squirmymcphee (talk) 19:09, 11 March 2008 (UTC)Reply
Finally got a chance to read the WaPo article. I was kind of surprised to read that they were disposing of silicon tetrachloride, but I guess that's explained by the desire to ramp up production quickly and at all costs. I feel like I must be missing something, though, when I read the quote on the cost of polysilicon production -- am I crazy, or does $84,500 per ton translate into $84.50/kg? That's way too high to be accurate.--Squirmymcphee (talk) 04:00, 12 March 2008 (UTC)Reply

Disregard Alcoa, I found the connection. Mr. Bingyan is a top dog with the Jinglong group and JA Solar. So there's a slight conflict of interest using him as a source in an article dealing with a competitor. Funny...

http://investing.businessweek.com/businessweek/research/stocks/people/person.asp?personId=31207526&capId=31160150&previousCapId=31160150&previousTitle=JA%20Solar%20Holdings%20Co.,%20Ltd.

I've tried to follow Rogol but he never seems to pop up on my google news searches. His prescient Sun screen studies hit the nail on the head four years ago. These last few years any amateur could throw a dart at a list of solar stocks and make money but Rogol was way ahead of the curve. I didn't realize he's forecast a slowdown. I'll have to follow up.

Here's a relatively recent article that mentions Si production costs for LDK. The numbers jive with the WAPO article. I was thinking the $84,500/ton was the cost to build the facility though. I don't have a good perspective on refining costs though so I'll have to follow up on this info too.

http://www.greentechmedia.com/articles/ldk-solar-cfo-talks-silicon-527.html Mrshaba (talk) 21:26, 12 March 2008 (UTC)Reply

I don't know that Rogol has forecast a slowdown, I was asking you whether you knew. Aside from not really having paid much attention to forecasts lately, Rogol has starting charging quite a bit for his reports and my employer hasn't purchased them, so I haven't really had a chance to follow his latest analysis.
The LDK article is confusing. They say their cost is $80/kg, but that the cost for incumbents is $30/kg. The $30/kg is consistent with what polysilicon manufacturers have always told me. The $80/kg cost strikes me as ridiculously high -- prior to 2006, I think you'd have to go back to the '70s not to lose your shirt on polysilicon at $80/kg. I don't have the impression that this article or the WaPo article were citing this as the cost to build the facility, but if I remember correctly (and I may not) that value may not be far off the mark for the typical cost to build a silicon refinery. Something about these articles doesn't make sense to me, but I just don't have time to put much thought into what at the moment.--Squirmymcphee (talk) 01:53, 13 March 2008 (UTC)Reply

My mistake... I checked out Photon Consulting. Judging from the synopsis Rogol et al. project demand and growth to be stronger than almost anyone imagines.

http://www.photon-consulting.com/en/solar_annual_2007/index.htm —Preceding unsigned comment added by 24.85.246.143 (talk) 18:31, 13 March 2008 (UTC)Reply

Solar energy page

edit

Hello Squirmy,

I did a lot of work on the Solar energy page last year but have not contributed for some time. After looking over the page again today I can see it needs some work. I have two questions you might be able to help with.

  1. 1: I read a book called Solar Power and Fuels a few months back. This book covered the first solar chemical conference back in the mid-1970s. The book defined solar chemical processes as chemical processes driven by photons. i.e. They specifically excluded processes driven by heat. Does this basic definition make sense to you?
  1. 2: When I think of the topic of Solar energy I divide it in two - I see solar energy as a resource on the one hand and solar energy technologies on the other. I've had a difficult time getting this distinction across but I think it's fundamental to the topic. Do you agree with this distinction or would you describe solar energy in an alternate way? Mrshaba (talk) 01:55, 10 April 2008 (UTC)Reply
I suppose that technically speaking a photon-driven process could be accurately described as a chemical process, but it isn't the sort of thing that immediately comes to mind when I read the phrase "chemical process." I generally think of a chemical process more or less as something characterized by an exchange of electrons without the involvement of other particles. I doubt that's a technically correct way to think about it, but even chemists frequently label chemical reactions involving photons as "photochemical." I'm not entirely sure what you're getting with your second question, perhaps confirming the difficulty you've been having -- are you trying to make a distinction between the solar energy resource and the tools employed to use it?--Squirmymcphee (talk) 02:46, 11 April 2008 (UTC)Reply
The chemical processes discussed in the book all involved electron exchanges but these exchanges were driven by light. By this reckoning a solar furnaces that cooks up a batch of lime doesn't qualify as solar chemical. One technology discussed was a hydrogen evolving photochemical process and there were other processes capable of growing labile chemical compounds that stored energy. I was hoping to improve the organization of the technologies on the solar energy page and I was wondering what your take on the solar chemical category was. Basically, I'm trying to draw a reasonable line between solar thermal and solar chemical processes. Hope that makes sense.
I'll change the second question to something hopefully less obscure. Do you think wind turbines, dams or downdraft towers are solar energy technologies? All of these utilize derivative forms of solar energy but I wouldn't call them solar technologies. On the other hand I think it makes perfect sense to talk about wind, hydroelectricity and warm tropical waters as secondary forms of solar energy from the resource point of view. Mrshaba (talk) 21:59, 11 April 2008 (UTC)Reply
That distinction between chemical- and heat-driven reactions seems sensible to me. On the second question, when you get right down to it virtually all of our energy (just about everything but nuclear) is a derivative form of solar energy. In my experience, though, when someone refers to solar technology they're generally referring to technology that uses solar energy directly.--Squirmymcphee (talk) 04:39, 13 April 2008 (UTC)Reply
I agree that most of our energy is a relative or a descendent of solar energy. I'm not asking you about that. I'm asking what do you think the term "solar energy" means to people. Do you associate the term with technology a resources or both? Mrshaba (talk) 07:41, 19 April 2008 (UTC)Reply
That depends on context, so I guess the answer is both, but usually not both at the same time. When discussing energy conversion, I think "solar energy" refers to technology that uses the solar resource as its primary input (i.e., without intermediate conversion to some other resource such as wind or tidal energy). Outside of that context I think it generally refers to resources alone, as in meteorology and biology, for example.--Squirmymcphee (talk) 05:23, 23 April 2008 (UTC)Reply
I agree it depends on context. It's a seemingly simple distinction but as I said above I've had a hard time getting across the two sides of the term "solar energy". I think the use of identical vocabulary to describe both the technology side and the resource side has a way of blurring the separation. Direct and indirect are used to describe types of solar water heaters, qualities of light and qualities of solar resources. I've tried to carefully choose words such as primary and secondary to replace direct and indirect when discussing solar resources but this replacement presents an added problem. To quote you, "...it's like a word that everybody understands -- it's okay to use a different word to represent the same idea, as long as you understand you might have to do a little extra work to make sure everybody understands your definition."
I should be happy we're no longer in the days of silicon batteries. Thanks for your reply. Mrshaba (talk) 17:10, 23 April 2008 (UTC)Reply
Hello Squirmy,
I've been working on the solar energy page lately. I was hoping you could give me your opinion of it. What is good and what is poor? Is it balanced and are all the notable topics hit on? Mrshaba (talk) 06:26, 28 April 2008 (UTC)Reply
Sorry, haven't had time to check it out lately and probably won't for the rest of this month. I'm not sure why I even bothered logging in today, actually.... I'll have a look at it when I get a chance.--Squirmymcphee (talk) 14:20, 7 May 2008 (UTC)Reply
No worries, I respect your opinion is all. I opened up the page to peer review so this should provide some positive feedback. Mrshaba (talk) 18:13, 7 May 2008 (UTC)Reply
Hello - there are more than two people working on the Solar energy article. Please keep discussions on the article in one place - on the Talk:Solar energy page. 199.125.109.41 (talk) 23:38, 10 April 2008 (UTC)Reply

Do you know and did you know

edit
  • Do you know of a recent estimate for the worldwide surface area of PV? I know it's a weird metric but I'm looking to make a comparison between PV and solar water heaters which are often measured by surface area.
  • The PV page is listing 2007 production as 3.8 GW and total installed capacity as 12.4 GW. The source is [1] At first I thought these numbers were much too high based on the info Maycock gave me last fall but just now I was reading some stuff from Photon Consulting that you might find interesting. [2] PC is listing installations as 2.6 GW in 2006 and 3.9 GW in 2007. Their projection for 2010 is stunning and I must say the report looks tantalizing. I suspect a lot of sources use Maycock's/Greentech's numbers so there's some scatter between reports that use info from different sides. Anyways... cheers Mrshaba (talk) 07:29, 24 May 2008 (UTC)Reply
    • Disregard... The Earth Policy Institute and Photon reports are listing yearly production (~3.9 GW) rather than yearly capacity additions (~3 GW). Apparently, there's getting to be quite a lag between the two. Mrshaba (talk) 18:08, 30 May 2008 (UTC)Reply
I'm looking for a nominal price on PV electricity for the solar energy page. I used 20-40 cents/kWh based on the solarbuzz site. Do you know of a better value? Mrshaba (talk) 20:31, 12 June 2008 (UTC)Reply
Been a rough couple of months, sorry to take so long. I think the Solarbuzz figure is probably good for residential PV. You might have a look at the DOE's Multi-Year Program Plan for 2007 -- I don't think it ever got past the proof stage and now they're working on something new, but it has some useful figures in it. Don't know if there are any good links on the NREL or EERE sites, so you'll probably have to Google it.--Squirmymcphee (talk) 21:19, 3 July 2008 (UTC)Reply

Claverton Energy Group

edit

Dear Squirmy - the above group may interest you.....Engineman...Engineman (talk) 06:13, 31 May 2008 (UTC) a lot of discussion on energy topics particularly intermittency....http://uk.groups.yahoo.com/group/claverton/ Are you aware of the sudden failure of Sizewell B and Longannet in the UK recently leading to a 1.5 GW loss of power in 10 minutes? Widespread power failures.Reply

Haven't had a chance to look yet. By chance does it discuss effective load carrying capability (ELCC)? That's a figure that is designed to take into account all of the effects of intermittency, usage patterns, etc. There are several recent studies on wind and at least one on PV that suggest their "equivalence" to a conventional power plant is highly dependent upon things like market penetration, local climate, etc.--Squirmymcphee (talk) 21:25, 3 July 2008 (UTC)Reply

Martin Green

edit

Hi Squirmy

How's it going? I was wondering if you could point me to the paper by Martin Green that you described as seminal? Are there any other must reads for those interested in PV? Mrshaba (talk) 22:16, 6 August 2008 (UTC)Reply

Going okay, been a really busy summer. I think this is the paper I have in mind, though it is not the first he ever wrote on the subject (I seem to recall a plenary paper at one of the major conferences in the late '90s, but I can't quite put my finger on it): Green, M.A., “Third Generation Photovoltaics: Ultra-high Conversion Efficiency at Low Cost”, Progress in Photovoltaics: Research and Applications, Vol 9, pp. 123-135, 2001.
The PV field is, in my opinion, surprisingly short on must-reads. You might check out Dick Swanson's "A Vision for Crystalline Silicon Photovoltaics" and the "companion" article by Stephen Hegedus on thin films, both published in the same special issue of Progress in Photovoltaics in 2006. Jozef Szlufcik and Johan Nijs have also written several very nice review articles on the state of manufacturable crystalline silicon PV technology, though I don't think they've written any since they left IMEC to form Photovoltech a few years ago. Ken Zweibel has written a few overview papers on thin films, though he has almost invariably turned out to be overly optimistic. And for a good overview of the physics of solar cells, both crystalline Si and thin film, I like Jenny Nelson's appropriately titled book "The Physics of Solar Cells." If I think of anything else I'll let you know.--Squirmymcphee (talk) 17:11, 20 August 2008 (UTC)Reply
Thanks for the info. I'm curious if you can answer another question regarding this article.
Here's the quote: "Polysilicon is the key raw material used to make solar cells, accounting for 70 percent of Suntech's costs of goods sold."
Do you think they're talking about raw material costs alone or are crystallization and sawing costs included in this 70 percent figure? Another way to put it is, what costs are generally included on the silicon side of things? Page 25 of this book has a PV module breakdown that groups wafer costs together on one side with cell fabrication and module assembly making up the remaining portion of module cost. The breakdown goes:
10% cell fabrication = $.30/Watt
25% module assembly = $.75/Watt
23% Poly-Si = $.69/Watt
23% Crystal growth = $.69/Watt
19% Sawing = $.57/Watt
I think this breakdown is surely out of date but it's a start. Do you know of an updated breakdown? Mrshaba (talk) 02:15, 23 August 2008 (UTC)Reply
This can be a bit tricky. For starters, the 70% figure for Suntech is undoubtedly the cost of a finished silicon wafer, not for raw polysilicon feedstock. Furthermore, if I'm not mistaken then Suntech is only shipping solar cells, not PV modules. Assuming that is the case, if you figure 25% of the cost of a module is for module assembly then silicon wafers account for 56% of the cost of the module. That sounds about right to me. If Suntech were buying finished wafers instead of making them from feedstock the cost would be a bit higher, since they would be paying not just for the production of the wafer but also for the ROI needed by the wafer manufacturer to cover capital investment and profit.
As for the breakdown from the PV handbook, if you bump the poly cost up to about $1.00/watt and cut each of the others by $0.10-0.15/watt you're probably in the right ballpark. A lot of these figures depend on process details, though. As I recall -- and I may well be wrong, as I didn't double-check -- this breakdown is for modules produced from single-crystal silicon solar cells. Crystal growth costs are less for multicrystalline silicon, sawing costs are the same or higher, and feedstock costs are generally the same or slightly lower. The total wafer cost is, in the end, lower, so cell fab and module assembly assume higher percentages of the total. As a rule of thumb I generally use 50-60% for silicon wafers, 15-20% for cell fab, and 30-40% for module assembly. I don't really have any direct references for that, it's just something I've worked out myself over the years. If you would like to see other breakdowns, though, look for papers by Jester et al. (generally attached to titles touting a Cz "renaissance"), Margadonna and Ferraza, and many of Maycock's reports (if you can get your hands on them -- I no longer have access myself).--Squirmymcphee (talk) 21:43, 27 August 2008 (UTC)Reply
I'm only trying for a ballpark understanding but this is tricky indeed... Thanks for the numbers but there are a few problems. Suntech is a module manufacture - the largest in the world. I agree the 70% seems to be for wafers but if that's the case I'm coming up with some odd percentages that I can't seem to reconcile. After looking over quoted remarks from the company that seem to be straight forward I've worked out an overall cost of about $2.33/Watt. This is broken into "non-silicon" costs of $.70 which I assume are for cell fab and module assembly; and $1.63 for the remaining "silicon" costs. This breakdown doesn't fit your thumb rules but much of this is likely due to high silicon costs. I'll have a look through the authors you suggested and see if I can't resolve the strange numbers I'm coming up with. I appreciate your insight as always.
One parting question, is it normal for wafer manufacturers to list their output in MW? Several manufacturers could use the same wafers as a feedstock but then produce modules that have a range of efficiencies. Is there a standard wafer rating factor I don't know about. Regards. Mrshaba (talk) 17:30, 28 August 2008 (UTC)Reply
Here's another one for you. I've been looking into wafering lately and it appears as though the equipment runs about $.45/watt. Does that seem right? I looked into one wafer manufacturer and surprisingly I found they were making about $.45/watt. Wow I thought... Going forward these high margins could be cut in half but the payback on equipment should remain quick. With the silicon bottle neck clearing up do you think the large module manufacturers are going to integrate wafering in the near future (next year or so)? Mrshaba (talk) 00:23, 30 August 2008 (UTC)Reply
I confirmed positively that the silicon costs from Suntech were for wafers. You were right from the get go. Rather than vertical integration I was told that "virtual integration" is the new game in town. This entails purchasing a minor share (10-20%) of upstream/downstream entities. I don't really buy this idea completely because it doesn't lead to enough control over pricing. But that's just me. You're the guy on the inside... What do you think? 24.85.246.143 (talk) 07:51, 1 September 2008 (UTC)Reply

Hey there... I've been digging into cost breakdowns for the last week. One thing I've found interesting is that many manufacturers seem to be moving towards using UMG in one way or another. Have you noticed that? This report mentions optimistic UMG costs from one manufacturer of $12/kg prior to directional solidification. The report also has a comparison table of manufacturing costs for poly/UMG plants which was something we went over above. Interesting stuff... Anyways... I still haven't figured out how wafer manufacturers can ship in MW when the wafer is basically a blank canvas that is used differently by different cell manufacturers. I'm also curious what is done with the silicon that is lost during sawing? Is it recoverable at all? Would you estimate sawing losses at about 2-3 g/watt? Are these sawing losses factored into the overall 10 g/watt figures that we talked about above. Mrshaba (talk) 01:31, 5 September 2008 (UTC)Reply

Hmm... I'll do my best to address everything since my last login all at once. First off, I know of no standard rating factor for wafers. Most likely whoever is quoting them that way is basing it on an assumed customer or industry-wide average efficiency. To me, it is akin to saying that a power plant will run X number of households -- it is approximately correct (though not always) and expresses the measurement in units that are directly comparable to something more common and accessible. To my knowledge it is press-release-type material, not something that anybody relies too heavily upon.
I was able to track down some specific figures for assumed efficiencies within SEC documents. They range from 14.5-16% and vary by manufacturer. It's a case by case situation. Mrshaba (talk) 03:34, 12 September 2008 (UTC)Reply
A cost of about $0.45/watt for wafering equipment doesn't sound too far off base to me, depending on the size of the plant. The $0.45/watt profit is almost certainly gross profit -- net profit will be significantly lower. A manufacturer will typically look for payback on the equipment of no more than a few years, and when equipment expenditures are high then a relatively high net profit is required to provide a reasonable return on investment to those who provided the capital. I think $0.45/watt gross profit is a bit more than most would require, but in the current market it wouldn't surprise me if somebody is getting that.
LDK is the company quoting $.45 gross profit per watt but others are quoting numbers closer to $.25/watt. There's a surprising amount of scattering between the wafer makers. I've repeatedly seen numbers around $.40-50/watt for the equipment. I'm still wondering how expensive this stage of the chain is. i.e. How do the costs of a $2.00 wafer breakdown between ingot, squaring, slicing etc. Some of the cell manufacturers are relatively transparent concerning their costs but I haven't been able to break apart the costs of making wafers with much success. Mrshaba (talk) 03:34, 12 September 2008 (UTC)Reply
How the cost of wafers breaks down is a bit variable, I think, and it's changing rapidly. Squaring is quite cheap, it's almost all in the ingot casting and the sawing. Then it depends on whether you're talking multicrystalline or Cz, how large an ingot you're casting, how thin you're slicing it, and how much kerf you're losing. About 10 years ago I remember being told that wafering was about twice the cost of casting, but I believe the cost of wafering has come down considerably with respect to that of casting since then -- partly because wire sawing technology has improved (it was pretty new in large-scale commercial use 10 years ago) and partly because wafers have gotten much thinner (more wafers from the same saw in roughly the same amount of time). I don't have any hard numbers, though.
I think there is plenty of interest in UMG, but I think it will be awhile before anybody is making or using it in large quantities. Solar cell manufacturers need to figure out how to use it without killing their efficiencies before any of them will buy it, so right now it's all talk. And if silicon prices collapse next year I expect that a lot of that talk will disappear....
We'll see... What I find interesting is that the two largest and many smaller cell manufacturers are using this stuff. I agree with your general observation though. Mrshaba (talk) 03:34, 12 September 2008 (UTC)Reply
Meh. UMG (and solar-grade, for that matter) don't have a standard definition, so without seeing the specifics on impurity content, etc., it's really hard to know what they're actually working with. Until the industry comes up with some standards they're just marketing buzzwords, as far as I'm concerned. Nearly all silicon feedstock is still made using the same processes and equipment used for microelectronics-grade -- they just cut a few corners here and there and voila! It's solar grade! That won't change anytime soon because the major feedstock producers want to hedge against a collapse in the PV market, so they make sure their new plants are capable of microelectronics-grade work. And the UMG stuff that seems to make the best solar cells actually goes through a surprising amount of refining, considering its moniker, because they have a lot of trouble getting boron and phosphorus content down to acceptable levels. Ultimately I think that somewhere in this direction is where the whole industry will eventually go, but I also think it's still a few years before its time.--Squirmymcphee (talk) 23:03, 18 September 2008 (UTC)Reply
An enormous amount of silicon is lost from in going from ingot to wafer. Anywhere from 20% to 35% of the ingot itself is lost, and about another 40-50% is lost during sawing. I think each of them works out to something like 2-3 g/watt, though the sum of those seems a bit high. Much of what is lost from the ingot can be recycled, but the sawdust -- the kerf -- currently cannot be. Several groups are working on that, though. And paradoxically, the thinner you slice your wafers, the more you lose to kerf (since the thickness of the removed layer remains about the same), but the lower your losses per watt become (since you produce more wafers and, therefore, more watts). Kerf and the discarded portions of the ingot are all included in the 10 g/watt figure.--Squirmymcphee (talk) 21:31, 11 September 2008 (UTC)Reply
This [company] recycles the SiC and glycol used in the wafering process and I wrote to them concerning the recycling of kerf dust. Can you list one of the groups working on recycling kerf? I find getting 160 kg of usable wafers from 290 kg of raw material to be a sinful amount of waste. Thanks for the explanations - especially the kerf loss question which I've made no headway on. Chieers. Mrshaba (talk) 03:34, 12 September 2008 (UTC)Reply
There's an EU-funded effort, though I forget what it is called -- you might google PVPS and see if it's one of their many projects (don't expect it to be any better organized than any other EU web site though). Solarworld has the biggest private effort that I am aware of, but it started with much fanfare and I have heard nothing since. The major issue is the amount of carbon (from the SiC) and metal (from the sawing wires) that ends up in the slurry with the kerf. Both of those are solar cell efficiency-killers and they must be removed, but how do you do that economically? It's a much tougher problem than it seems.--Squirmymcphee (talk) 23:03, 18 September 2008 (UTC)Reply
Thanks for the info. SiC Processing confirmed they simply sell off the kerf/SiC product because recycling is not technically feasible. I'll have to read through those sources - Solarworld has a rather complicated website so that should be fun. Here's another one for you: Cell producers commonly list wafer costs for 5 or 6 inch wafers. Is there a standard way to translate a 5 inch wafer into an expected peak-watt at the cell or module end? I suppose I could look at the area of a 6 inch wafer and use standard test conditions with average efficiency to get watts. The problem with this approach is that I assume there's some squaring of the wafer that's done so I have to return to the original question. Is there a standard way to translate 4/5/6 inch wafers into expected watts? Mrshaba (talk) 05:44, 19 September 2008 (UTC)Reply
The specified size is for the finished wafer, so it takes any squaring into account. But then you still need to look at the shape of the wafers. If you're talking about square, multi-crystalline wafers then just look at the area and do the math. But if you're talking about the pseudo-squares with rounded edges that are typically produced from single-crystal silicon then it is more complicated. There, they have a long round boule after production, which then is shaped (i.e., has the sides shaved until it is the right diameter), then squared, then cut into wafers. To calculate the area you need to know the radius after shaping so you can calculate the area missing at the corners of the pseudo-square. They are standard sizes, to my knowledge, but I don't know offhand what those standard sizes are. However, I can tell you that for a 6-inch pseudo-square the area is 236.5 cm2.--Squirmymcphee (talk) 10:00, 21 September 2008 (UTC)Reply

I am a little amused to hear you talking about round wafers that are 4-6 inches in diameter (100 to 150 mm). 30 years ago the semiconductor industry was just switching from 3 inch wafers to 4 inch wafers, but I think that today 12 inch diameter wafers are the norm (300 mm). I just saw a photo somewhere of a rectangular block of silicon that was made for solar panels, measuring maybe 150 mm by 300 mm by 400 mm (6in x 12in x 15in). However, I am not sure if it was poly or single crystalline. Poly I would guess, as the, and I am not going to try to spell it cz something process involves pulling silicon ingots while rotating them to create a single crystal, making them inherently round. Apteva (talk) 01:57, 23 September 2008 (UTC)Reply

Yes, generally when you see a rectangular block of silicon it will be cast multicrystalline, while single-crystal will generally be found in cylinders (or cylinders with the sides shaved mostly flat) prior to wafering. The dimensions you give for the rectangular block are a bit odd -- I would expect it to be square across one plane -- but the block was almost certainly cut from a much larger ingot. The norm nowadays is about 690 mm by 690 mm by 200-250 mm, though equipment is available to do ingots about twice that volume. Those ingots are chopped into smaller bricks, usually 100-150 mm in two dimensions and 180-220 mm in the third, then sliced into wafers. You will typically get 16 or 25 bricks from each ingot, though the largest ingots may yield 36. At any rate, there is a very simple reason the PV industry sticks to 4-6 inch wafers: Solar cells need to be series-connected in long-ish strings to produce useful amounts of voltage. If your cells are 300 mm in diameter and you need enough voltage to charge a lead-acid battery then your module will be nearly 2 m by 2 m (and the increasingly popular higher-voltage panels would be more than 2.5 m by 2.5 m). Such a module would be quite unwieldy to handle during assembly, shipping, and installation. It may also require slightly thicker glass -- I'm not clear on that -- which is already the most expensive material in the panel behind silicon. From the cell manufacturer's point of view there would be a lot of upside to using larger wafers, but it is more than outweighed by the downside on the module end. Right now the conventional wisdom seems to be that the optimum is in the 150-200 mm range, with most betting on (or at least settling for) the low end of the range. Sharp has experimented with 200 mm wafers, but the reaction the results provoked was along the lines of, "I'm glad they did that, because now we don't have to."--Squirmymcphee (talk) 17:05, 23 September 2008 (UTC)Reply
It could have been 690 mm. All I know is that it was bigger than 6 inches, but possibly less than 6 inches thick. It could have been square. I would think that it would be helpful to put into the photocell article a list of common cell sizes. Intuitively if you are covering a football pitch with cells you don't want them to be 4 in dia. Or even 6. However it does seem that half the market is small installations and half large ones. Apteva (talk) 03:54, 24 September 2008 (UTC)Reply
I am preparing some edits for the solar cell article, and perhaps I'll add standard cell sizes while I am at it. As for covering a football pitch, system installers really couldn't care less about the size of the cells. They just want modules that produce the voltage and current required for their project, are large enough to minimize repetitive installation tasks (and therefore labor costs), but are small enough to handle easily (thereby avoiding extra shipping, handling, and labor costs). The advantage of large cells comes during the cell manufacturing process -- double the area of the wafer and you double the number of MWp your plant produces without changing the number of wafers you handle. So cell manufacturers would love to make bigger cells, but for most projects, system installers don't want the larger modules that result. I think cells will get bigger than they currently are, and in fact just yesterday I read that one manufacturer is targeting 210 mm cells in the reasonably near future. How big they get depends on how many customers want modules bigger than a king-sized bed. Any expansion in wafer size will be done cautiously, though, because for most (if not all) PV companies switching to a larger wafer size means building out a brand-new production line, so it cannot easily be done on a trial basis.--Squirmymcphee (talk) 15:03, 25 September 2008 (UTC)Reply

Solar Energy Article

edit

Thanks for stopping by, please come by frequently. --Skyemoor (talk) 19:43, 15 October 2008 (UTC)Reply

Current Affairs

edit

What's your take on the current situation for the PV market? I can't believe how far some of these companies have fallen. Consolidation? Die off? It would seem the German producers will get nailed the hardest unless Germany enacts import restrictions. Mrshaba (talk) 03:51, 28 October 2008 (UTC)Reply

All of the markets for everything have been dropping lately, so I haven't been reading much into the drop in the PV market in particular. It certainly seems to have dropped more than most markets, but I think the biggest threat is to capital-dependent startup companies that have no product to sell. PV companies with a product to sell (and even some without, e.g., Suniva) are sold out well into 2010, and it would take a lot of cancelled orders to change that (they oversell in much the same way airlines do). I think I've told you before that I believe consolidation will happen someday, and the current financial crises might accelerate that, but thus far I have neither seen nor heard anything to suggest that something is imminent. (There are rumors that somebody wants to buy Evergreen Solar, but there are always rumors that somebody wants to buy Evergreen....)
As for Germany, to be honest I haven't paid much attention to the German market lately and I don't really understand the reasoning behind your statement.--Squirmymcphee (talk) 22:39, 3 November 2008 (UTC)Reply
As I understand it, the Chinese producers have a sizable cost advantages over the German manufacturers. The pundits are forecasting an oversupply issue next year and a 10-20% drop in prices. It would seem that falling prices will hurt the German manufacturers more because a higher percentage of their profits will be effected. It's speculation on my part but a solar tariff would tend to counter the Chinese production advantage. Mrshaba (talk) 05:16, 4 November 2008 (UTC)Reply
A lot depends on how the various companies have set themselves up financially. Companies that have committed themselves to buy feedstock or wafers at high prices for the next several years are going to have bigger problems than those who have some flexibility in their contracts, for example. It will also depend on how their lines are set up on how reliant they are on manual labor (on cell production lines at large manufacturers I think you will find little reliance on manual labor in either country). But regardless, a tariff would push PV prices up just as they're finally beginning to come down. I don't that's a good idea. I also don't think it would have the desired effect. I don't have any figures on how much PV the US imports from China, but some 3.5% of US PV exports go to China. Furthermore, the US is still a net exporter of PV modules so for a tariff to have the desired effect I think it would have to be levied by some other government that is more significant to the Chinese PV industry.--Squirmymcphee (talk) 11:55, 10 November 2008 (UTC)Reply
To be clear I was talking about a German tariff on Chinese imports but I guess the strengthening dollar already acts like a tariff because it's hitting all the manufacturers shipping to Europe who report earnings in USD. Nearly all of the Chinese manufacturers have been shipping 66-75% of their product to Europe. This report says 60% of Chinese product was shipped to Europe in 2007 but factoring in the Spanish market this last year will push the percentage higher. I don't think 3.5% of US exports are going to China. If I had to guess I'd say that number represents cell shipments that are packaged in China - Sunpower outsources some module fabrication for example.
As I see it, the strongest PV companies have been the one's locking in silicon supply so I think the weakening effect that these contracts will have is being exaggerated. I haven't read the contracts myself but the news reports generally mentioned that there was either feedback between spot silicon and contract prices or a declining cost curve - falling silicon prices have been forecast for a long time now after all. We'll see... Mrshaba (talk) 21:42, 17 November 2008 (UTC)Reply
The strengthening dollar definitely has an effect on Chinese manufacturers because their currency doesn't float -- it is pegged to the dollar. Last week I noticed doom-and-gloomish press releases from several Chinese PV companies, but did not see anything similar from European ones; on the contrary, some of them expressed cautious optimism, and Solarworld even decided it could afford a billion euros to buy Opel, if GM is willing to sell it. The Europeans do not seem too worried about China at the moment.
I was off just a bit on the export figure -- it's 3.4% of US exports that go to China. At least, that's what US PV manufacturers reported to DOE in 2006. It doesn't differentiate between cell and module shipments, so it is quite likely that at least some of that was cells that were packaged in China and possibly re-exported.
In the near future I think everybody is going to be able to lock in silicon supply and spot market prices will collapse almost to the level of contract prices (which will themselves decline). There is without question a changing price schedule built into silicon supply contracts, though I can't say how closely these schedules are tied to the spot market price. If the market is flooded with as much silicon as some are speculating then I won't be surprised if the spot market price drops by a factor of five or more in the next 6-12 months, but that's not going to happen to contract prices.--Squirmymcphee (talk) 18:41, 22 November 2008 (UTC)Reply

Solar Fraud in Spain

edit

Here's an interesting story you may or may not have seen yet. [3]

Here's a question for you. If you can use microjet lasers to produce the grooves for buried contacts and slice wafers why aren't they used in processing whole ingots? i.e. Why not use lasers for blocking and wafering? Mrshaba (talk) 05:52, 16 December 2008 (UTC)Reply

I usually keep up with Greentech Media at work via their daily emails, but I've been on vacation since the 13th so I hadn't seen that one yet. The Spanish market is a mess right now. I understand what the government there is trying to do, but they've done a terrible job of it and have effectively eliminated one of the world's largest PV markets for the coming year....
Laser cutting is a bit out of my area, but as I recall you can have big problems with thermal erosion inside the grooves when making deep cuts with a laser. Thus, a wafer sliced with a laser is likely to have uneven thickness. Since you specifically refer to microjet lasers I presume you're thinking of the ones where the laser beam is guided by a water jet, which would presumably cool the surfaces inside the groove and perhaps remove this barrier. To my knowledge nobody has ever demonstrated blocking or wafering with such a laser, which is certainly a prerequisite for using it production. The company that has been pushing that technology has demonstrated it for cutting EFG ribbons into wafers and dicing microelectronic devices, but not for cutting through entire silicon blocks or ingots. I imagine it would be very difficult to get the water to continue into a deep groove without losing its shape and disrupting, instead of guiding, the laser beam, but that is really only an educated guess on my part.
For blocking the answer is much easier: band saws are much cheaper than lasers, fast, and produce very good results.--Squirmymcphee (talk) 14:03, 19 December 2008 (UTC)Reply
I hear what you're saying about thermal effects and the difficulty of maintaining laminar fluid flow but it appears they can already go 7 cm deep which would mean diameters up to 14 cm are achievable. Here's a paper I think you'll find interesting and educational. Apparently the quality of wafers produced with the liquid-jet guided lasers is good enough that the etching step is reduced or eliminated. Cutting speed is higher and kerf loss is reduced. The paper doesn't say if kerf recycling is possible but that also seems to be a possibility depending on the chemical composition of the jet. No mention of cost unfortunately - Got any contacts a Fraunhofer? Mrshaba (talk) 17:37, 19 December 2008 (UTC)Reply
Question: What is the standard way to measure ingot costs? Mrshaba (talk) 06:16, 23 December 2008 (UTC)Reply
So... I guess you don't buy the laser wafering idea? I don't know myself but it's interesting. Here's another laser idea for you. Have you heard of atomic vapor laser isotopic separation (AVLIS)? The basic idea is that you tune a laser such that it ionizes a specific molecule that is then removed with electrostatic precipitators. The technology has been used to enrich uranium and lead among other things. I was thinking the same process could be used to separate impurities from silicon. I've looked over the separation costs for uranium and it seems AVLIS could be competitive with the standard silicon refining technologies. Have you ever heard of using something similar to AVLIS to purify silicon? Mrshaba (talk) 08:18, 28 December 2008 (UTC)Reply
Here's a laser purification device. This laser technique proceeds by pyrolyzing the target impurities in silane. Mrshaba (talk) 20:30, 9 January 2009 (UTC)Reply
Interesting paper -- I had not seen that one. It looks promising, though rotating a brick 180 degrees to cut from the other side would almost certainly require an etching step to reduce the inevitable surface nonuniformity. The etching advantage would only be an advantage to some customers, though -- for many manufacturers, surface texturing doubles as the damage etch and texturing would still be desirable. I really wonder how much it would cost, though if it can be done in production quantities as fast as that paper claims then they could be quite expensive and still break even with wire saws (though of course they need to have some advantage over wire saws before anybody will take a chance on them...).
Cost per kilogram is the only "standard" way I know to measure ingot costs, but that will be affected by the size of the ingot, the growth technique, and the cost of the silicon feedstock it uses. It would be really hard to compare on equal footing, say, an mc-Si ingot grown by BP Solar to one grown by REC Scanwafer (and even harder to compare to a Cz-Si ingot grown by SolarWorld or SEH).
Never heard of AVLIS. If I read you correctly, you have to have lasers tuned individually to each of the impurities you're trying to eliminate. That sounds really, really difficult unless you've already done some purification and you're simply targeting a few known impurities. Still, I suppose it's possible there is a sequence of purification steps in which it might make some sense. My gut feeling is that you won't see much change in the way silicon is purified until the solar cell manufacturers themselves get big enough to vertically integrate to that step (it is already beginning to happen with the silane-to-feedstock step). Right now, feedstock manufacturers are afraid to build plants they can't use for electronics-grade silicon in case the bottom drops out of the PV market, and I think that's the way things will continue until they're forced to compete with other manufacturing methods.--Squirmymcphee (talk) 14:19, 18 January 2009 (UTC)Reply
Hey Squirmy... I'm glad you liked the paper. I wrote to one of the authors asking about the kerf "recyclability" of the process but I didn't get any word back.
You wouldn't want to use an AVLIS like technique on a raw feedstock but it might be helpful to remove the last 5 ppm of boron. Ideally you'd want to get Boron in compounds like B2H6 or B2O3 to simplify the task but I don't know how this would be done. There's a limited amount of info available so I couldn't make any meaningful comparisons.
Have you run across this paper by Severin Borenstein. He derives a time weighted value (10-12 cents/kWh) to centrally generated PV electricity. He compares this to the costs of generating the electricity (40 cents/kWh) and concludes #1. Central PV is unlikely to become competitive #2. We (California) are throwing money away on subsidies and should wait until the technology gets cheaper. #3 The value of PV to end-users is much higher and this is where it should be going. There's a video that covers the paper here. I find these results thought provoking. Do you debate the value of subsidies with your colleagues? Mrshaba (talk) 20:16, 18 January 2009 (UTC)Reply
Seems to me that to get boron into compounds like B2H6 or B2O3 you would need to get your whole mix of stuff into gaseous form, as those compounds do not exist as such in liquid or solid silicon. Not so sure that would be useful for the last 5 ppm of boron....
The Borenstein paper looks interesting. I had not seen it. I have now downloaded it, but only skimmed it. His levelized electricity cost number of 40 cents/kWh for $8/Wp installed looks good, provided he's talking about a residential PV system funded with a personal loan. Few banks will loan enough money to pay for a PV system without having real property as collateral, though, and then you're talking about mortgages and the like, where the interest is tax-deductible. That alone will cut the levelized cost by about 20%. Then if you want to talk about commercial systems you get additional tax write-offs for depreciation; depending on the situation, there may also be state and federal investment tax credits. At any rate, you're then talking about 18-20 cents/kWh. I'm not talking about renewable energy subsidies here, I'm talking about the same benefits that are available to anybody who makes a capital investment -- the same thing you would get if you borrowed against your home equity to build a new kitchen, or that a construction company would get for buying a new crane. Borenstein doesn't take any of that into account, and they are such significant factors that I think it completely ruins his analysis. At this point, his $8/Wp installed figure strikes me as high as well, though at the time he did his analysis it may well have been accurate. According to the numbers I just threw at you, that puts central PV within spitting distance of Borenstein's valuation of centrally generated PV. (The numbers I'm using, by the way, come from DOE and are calculated almost exactly the same way that Borenstein did his, but with allowances for tax-deductible interest, depreciation, etc.)
Where subsidies are concerned, I have mixed feelings. There is a lot you can say about the poor economy and the ending feedstock shortage that might explain falling PV module prices, but balance-of-systems costs seem to have dropped like a rock over the past 6-18 months. I don't have any hard data on that, but I have seen and heard quite a lot of anecdotal evidence to that effect. I firmly believe that this, if true, is largely a result of capital investment in PV installers, hardware, and related equipment (inverters, etc.), none of which would have happened without subsidies. Borenstein discounts learning-by-doing as a factor in the decline of PV prices, but all of the studies he cites in support of that conclusion consider only the PV module, which is only half of the installed system cost. I am not aware of any learning curve studies like Nemet's that attempt to assign specific causes to price declines in BOS components, but the fact remains that Borenstein makes his conclusion on the basis of only half of the required information.
Given all of this, I disagree with conclusion #1 (as you have labeled them). I think most PV will ultimately be distributed, as that is one of PV's strengths, but I do not think central-station PV will be a rarity. Plus pricewise, it is a lot closer to competitiveness than Borenstein thinks. Were his cost calculations correct I might agree with #2, but instead I'm on the fence. My understanding is that California's subsidies are widely expected to have a net long-term benefit to ratepayers there, but it isn't something I've looked into; if true, then I have to disagree with #2 as well. As for #3, once you correct Borenstein's math I think you'd find that whether this statement is true will depend very much on the cost of peak power generation (and therefore on fuel prices).--Squirmymcphee (talk) 20:13, 29 January 2009 (UTC)Reply

Hmmm... Gaseous you say? Technically, that's ok because AVLIS works better on gases than liquids but I don't know if that's economically feasible. There's been AVLIS type work in the liquid phase but it's trickier (as I understand it) because, among other things, your ionized particles don't travel as far or as fast. I was thinking you'd only use a laser separation step as part of a larger purification chain (vacuum distillation, laser separation, directional solidification). As I hope you can tell, it's just a hair brained idea and I don't want to beat a dead horse.

Jeez.. Thanks for blowing Borenstein out of the water. I was getting to like the guy. I checked his installed costs against the CSI database and they were roughly correct. I also read some of the critiques of the paper but I guess I missed the obvious omissions that Borenstein made.

As far as subsidies go, I have mixed feelings too but it seems obvious that subsidies have lead to scale and scale has lead to falling costs/prices. By this reasoning I think Borenstein mischaracterizes Nemet from the get go. I might be missing something but that's my take. Overall though, I'd like to see subsidies phased out as quickly as possible.

Here's another paper by Borenstein that talks about real-time pricing (RTP) and how it's a more economically efficient way to run the grid. You end up using about the same amount of electricity but less idle capacity is required so you have a 5-10% net savings to the entire grid. Pacific Gas & Electric and other utilities are moving to Smart Meters throughout their service territories. The motivation for this is workforce reduction but an ancillary benefit is the ability to meter by time of use. I had heard that TOU pricing penalized PV users but Borestein finds that 95% of PV owners would benefit under TOU pricing. The other interesting thing to think about is that RTP drives all sorts of demand response that you don't get with fixed rate pricing. When you look it from this angle you can see that demand response behaves as a sort of back-up power sink for intermittent power sources. Mrshaba (talk) 23:17, 30 January 2009 (UTC)Reply

My understanding has always been that TOU benefits PV owners since solar arrays tend to generate electricity when it is most valuable. However, I have never had the data to test that myself.--Squirmymcphee (talk) 22:00, 11 February 2009 (UTC)Reply
I think there's a lot of misunderstanding around the benefits of real time pricing (RTP) and time of use. The Borenstein study runs the numbers in a legitimate way and mathematically shows that it's a good idea for PV users. Unfortunately, the CPUC was very quick to revoke the TOU requirement following the LA Times article. What puzzles me is the failure of the PV community to promote RTP and TOU?
I don't know why kerf loss bothers me so much but here's an article you might have access to. I think I've heard you say before that poly prices will come down far enough that this won't matter. I suppose you're correct for the short term but you'd hope that some sort of recovery process can be developed going forward. I happened across the kerf recovery article by chance - I was actually googling the idea of centrifugal silicon purification. Ever hear of Delaval purifiers? Same basic idea. I'm surprised I haven't found any mention of it in my searches so far. Mrshaba (talk) 17:26, 20 February 2009 (UTC)Reply
I wouldn't say that kerf loss won't matter, but the economic benefits of things like thinner wafers, UMG silicon, and kerf recovery will become so much smaller that there will be much less pressure to make them succeed (and much less profitability to motivate research and investment). It all comes down to the cost of recovering kerf -- with short silicon supplies and high prices, a lot of people would be interested, but with cheap, plentiful silicon it is much harder to recover kerf at a competitive price. Regarding your literature search, I have seen a couple of articles in the past about using centrifuges to recover silicon carbide particles for recycling (as I recall, they were published back when silicon was so cheap that kerf recovery wasn't given a second thought), so you might happen into some additional information if you search down that path.--Squirmymcphee (talk) 21:00, 26 February 2009 (UTC)Reply

Wafer $/Watt

edit

Do you have a line on what current wafer prices are and where they're headed in the next year or two? I came across this [4] today - I'm surprised by the guesstimates that have wafer costs coming down to $.85/Watt. Does this price seem reasonable to you? If you add this price to the non-wafer costs of 60-70 cents/Watt you get about $1.50/Watt (cost) about $1.80/Watt (factory gate price) which seems low to me. I think a panel price below $2/Watt is a great thing but there could be some unexpected problems on the regulatory/subsidy end. The mantra has always been that the panels represent about half of the installed system - will this remain true or will installation cost stay high? i.e. the cheaper panels prices won't be passed along to consumers. The problem, as I see it, is that the subsidy arrangements require that certified installers be used to qualify for the subsidy. This forms a road block of sorts. It seems more rational to have a performance based incentive that rewards results irregardless of certification. What do you think? Mrshaba (talk) 13:49, 7 April 2009 (UTC)Reply

This is very timely -- just this past week I have had similar conversations both at work and with the CTO at another PV company. The Barron's thing is by far the most dismal forecast I've seen. I can see wafer prices coming down to $0.85/Wp for higher efficiency multicrystalline cells. I don't see that as an industry-wide ASP, though I can see that dropping below $1.00/Wp. If he forecasts a 60% drop in polysilicon prices resulting in $50/kg then I suspect he's talking about spot market prices, and that kind of drop on the spot market has not been predicted in anything else I've seen. A couple of other analysts have the spot market holding steady in the $100-150/kg range, which they suggest indicates we are not yet in a silicon oversupply condition. Contract prices on polysilicon are definitely dropping, though, and that is how wafer prices will get so low. Polysilicon is the part of the supply chain that has the most to give right now -- the selling price can drop a lot while still remaining profitable -- which is why, I think, cost reductions are expected to be concentrated there.
The speculation I've heard on sub-$2.00/Wp modules involves dumping by Chinese manufacturers who have over-produced; if it happens, it will definitely be interesting, but in that case I would expect module prices to increase again after that. I think the $1.80/Wp factory gate price for modules is a best-case scenario -- somebody might attain it soon, but I don't see it as an industry average this year or next. That would, in fact, suggest financial losses this year and something like break-even next year, but I also wonder how these losses will spread through the supply chain (i.e., will module manufacturers take the brunt of it?). The problem some cell manufacturers are having now is that they have inventory that is technically sold, but the customer is refusing to take delivery or pay.
On total system cost, in the long run I think everybody expects the more-or-less 50/50 split between module and BOS to remain (maybe it will be 40/60, but still somewhere in the neighborhood of 50/50). In the short term, costs in both areas are changing so rapidly that it is hard to know what to expect for a system installed, say, a few months from now. I think lower panel prices will be passed on to consumers because they will have to be -- it's not like the last 10 years where an installer can give a price and say "take it or leave it," knowing that he'll find a customer who will take it.
On subsidies, I definitely think results should be rewarded rather than the simple fact that a PV system was installed, which is why I like feed-in tariffs. The certified installer thing doesn't bother me -- it's easy enough for a qualified installer to get certified, and I think we ought to be careful about incentivizing installations by people who don't know what they're doing. Perhaps the requirement could be that the system must pass an inspection against all applicable codes, but in that case you're still talking about 99% of installations being performed by somebody with the qualifications of a certified installer (or at the very least a professional electrician with a bit of knowledge about PV). DIY installations, if that's what you're thinking of, would still be limited to a few very knowledgeable amateur electricians.--Squirmymcphee (talk) 09:30, 11 April 2009 (UTC)Reply
Strange times these. Your logic on lower prices being passed on to consumers is sound but I'd like to see some proof of it. Disregard my grievance concerning installations by do-it-yourselfers. I was just now looking through the CSI documentation and it appears self-installations are allowed "so long as proper building permits are obtained and local codes are followed." I thought certified installers had to be used to qualify for the rebate but this doesn't appear to be the case. Mrshaba (talk) 17:09, 11 April 2009 (UTC)Reply

BOS $/Watt

edit

Hello, I ran into this [5] the other day. The report notes BOS costs/watt have declined over the last few years but I was disappointed that the analysis didn't cover 2008. You mentioned earlier that there's been a recent downward shift in BOS costs (I assumed you meant 2008). Can you suggest a source for this info? Mrshaba (talk) 05:27, 22 April 2009 (UTC)Reply

Not really -- my own source is just industry chatter. The decline in demand this year is causing the prices of everything to drop, but I don't know anybody who has a good handle on which items have dropped how much and why. Even if you talk to different installers you often get conflicting opinions. The only thing everybody seem to agree on is that prices are declining. When I make calculations for my own purposes I have been using something along the lines of $3.00-3.50/Wp for BOS, but that number is really just a semi-educated guess....--Squirmymcphee (talk) 16:27, 26 April 2009 (UTC)Reply
$3.00-3.50 seems fair. The LBL report placed average BOS costs at about $3.75 at the end of 2007. Going forward it will be interesting to follow the trend of qualified installers vs. BOS prices. Mrshaba (talk) 21:41, 28 April 2009 (UTC)Reply
Off the wall question for you. How low do you think the price of PV electricity can get? I'm looking for a cents/kWh estimate. I read a few interesting things in First Solar's recent earnings call. First, they mentioned that they are starting to worry about glass costs becoming a limiting cost reduction factor. I believe you've mentioned this before and that's why it jumped out at me. The other thing the earnings call mentioned was that there's a steep efficiency penalty on the BOS side. Here's the quote:
"...adjusting for a balance of system penalty of $0.15 a watt, which is what you get assuming 12% conversion efficiency on First Solar modules versus 14% on polycrystalline silicon..."
My guess is that most polycrystalline panels manufacturers will eventually reach efficiencies more along the lines off 16-18%. This means that they have an estimated $.30-.45/watt advantage on the BOS side i.e. poly-Si panels at $1/watt are about equal to thin films selling at $.65/watt. To be fair, there are lots of additional factors that come into play that I'm neglecting. The rough point I'm trying to make is that thin-films seem to take a serious penalty due to their low efficiency. My thinking is that the only way thin-film are guaranteed long term success is if it can improve efficiency significantly. What do you think? Mrshaba (talk) 06:22, 9 May 2009 (UTC)Reply
Here's a quick read you might enjoy. [[6]] Mrshaba (talk) 02:26, 11 May 2009 (UTC)Reply
A lot of people are thinking along the same lines as you. There were some papers by Georgia Tech, GE, and Applied Materials on the subject at the recent ASES conference in Buffalo, though I don't know whether they are publicly available at the moment (I don't have them). I imagine we'll hear a lot more about this in the coming year or so.--Squirmymcphee (talk) 05:38, 19 May 2009 (UTC)Reply
Ahhh... You're teasing me. I'm just a poor devil of a sub-sub looking for a whale bone... I will search for the papers... If I can get through Moby Dick I can find those papers - public availability be damned. Cheers Squirmymcphee... If you're up for exchanging emails I'd be delighted. No worries if anonymity continues... It's funny you know. I've spoken/written to so many wonderful characters in the PV pantheon but I enjoy our Q&A the most. It has a post card quality - a sort of PV letter chess. Mrshaba (talk) 04:40, 20 May 2009 (UTC)Reply
I'm not trying to be mysterious, I just don't know how long it takes ASES to publish its conference proceedings. It's really tough to pin down any sort of cost advantage, though, because it is sensitive to so many things. Ask Unisolar about what First Solar says and they'll tell you that with their peel-and-stick module technology BOS costs are so much lower that they are as competitive as First Solar (though I don't believe that). The point being that somebody will complain about your calculations if you put everybody on equal footing, but somebody will also complain if you give somebody else special treatment, and in the end what it all comes down to is what is possible on each specific site where PV will be installed. It's a tough nut to crack, but based on my conversations with various people we will some attempts to crack it in the coming year. In other words, I'm not sure there is a whale bone out there for you just yet, but if you're patient I think you might be rewarded.
Glad you enjoy our chats. I do too, which is probably why I have gotten so little editing done. Still, I prefer to keep it to this forum for the time being if you don't mind.--Squirmymcphee (talk) 13:53, 21 May 2009 (UTC)Reply

I couldn't find those papers and I needed a fix so I read through some earnings transcripts instead (LDK, Renesola, Suntech). Here are some of the details I found interesting.

-Suntech lists non-silicon costs at 66 cents/watt. They are at low utilization currently and hope for non-Si costs of 50 cent/watt down the road.
-ReneSola lists wafer processing costs at 36 cents/watt. (I had no idea processing costs were this low)
-ReneSola lists silicon utilization at 6 grams/watt (This value also seemed low to me.)
-ReneSola says ASPs have fallen from $2.16/watt (Q4) to $1.27 (Q1)
-LDK says ASPs have fallen from $2.18/Watt to $1.54/Watt
-LDK's processing costs are 30-40 cents/Watt with silicon costs close to $150/kg - total costs are currently right around their ASP of $1.54/Watt.
-LDK says they've started producing silicon with their own 1000 MT reactor and the 5000 MT reactor will come online shortly. Projected production costs are around $30/kg.
-Taken alone I wouldn't believe the LDK numbers or the ReneSola numbers but they channel check.

From the reports it seems likely wafer ASPs will fall below $1/Watt by the end of 2009. That's surprising by itself but then I look at where LDK is headed. Near term it looks like they will produce silicon for $30-40/kg and that means they can probably get near term costs down to about 50 cents per watt (hypothetical: 5.5 grams/watt, 3.5 cents/gram, processing costs of 30 cents/watt). I don't know how long it will take to get wafer ASPs down to 50 cents/Watt but the numbers clearly indicate this is where things are headed. When I put Suntech's hypothetical future non-silicon costs together with these hypothetical future wafer costs I come up with the magic $1/watt. Pretty neat.

I read an Applied Materials report today that had this quote: "Czochralski process intrinsically limited to ~200 mm (for solar)and, perhaps, 300 mm for ICs." Do you know why solar Cz ingots are limited to 200 mm?
This report [7] mentions that, "Proponents of 300 millimeter fabs have estimated the cost of fabbing wafers is 30% cheaper in a 12’ (300 mm) fab than in an 8” (200 mm) fab." When I read the note I assumed the processing costs for 300 mm solar wafers would also be reduced but this is only a guess. The contrast between the two quotes puzzles me? Any thoughts on this? Mrshaba (talk) 14:03, 29 May 2009 (UTC)Reply
Since you have to put cells in series to build up voltage, larger cells generally mean larger modules. At some point the module becomes too large to be manageable by very customers, so a dedicated 12" cell line doesn't makes sense to many people right now. I could see it happening at some point in the future if a company is convinced enough that it will be making large modules for utility-scale applications (where using heavy equipment to help move modules might be more practical at low cost), but right now I doubt you will see anybody commit to, say, 250 MW worth of 12" cells (50 MW/yr over 5 years).
Sharp has made 12" cells on a pilot scale and they worked fine, but the modules were huge. Some have proposed cutting the wafers in half to keep module sizes manageable, but I suspect that would require near-perfect yield not to erase the savings gained by going to 12" wafers in the first place. Inverters that can work efficiently with lower input voltages might also help. I wouldn't go as far as Applied Materials and say Cz is intrinsically limited to 8" for PV, but going much larger than that certainly increases the possibility of increased downstream costs that can wipe out your savings.--Squirmymcphee (talk) 12:07, 30 May 2009 (UTC)Reply
I hear what you're saying but wouldn't there be a considerable fabrication advantage moving from 200 mm to 300 mm wafers because you'd have half of everything to do? That seems like a huge advantage. I understand the upstream vs. downstream tradeoff. If you're worried about voltage why not integrate a small inverter/transformer into each 2x4 panel? Or wire your panels in series to get the desired voltage? You say there's a problem with the efficiency of low input voltage inverters currently? I'll have to look into this. Can you suggest a source?
I like the idea of integrating inverter/transformers into each panel. 1. Minimizes shading problems 2. Allows system owners to incrementally add capacity. 3. Removes an installation step. 4. Allows incremental replacement of inverters as they go bad. 5. No DC radio problems 6. I assume small inverters would move down the learning curve more quickly. Mrshaba (talk) 21:53, 30 May 2009 (UTC)Reply
For a grid-tied system, the voltage issue is one of efficiency. Inverters have a DC-DC converter inside them which is where most of the conversion losses happen, and boost converters (which increase the voltage) are less efficient than buck converters (which reduce the voltage). Thus, for a 120 VAC system it is preferable to have more than 120 volts on the DC side. That said, microinverters overcome panel-to-panel mismatch losses, and I think they have gone a long way toward minimizing or eliminating the efficiency gap between buck- and boost-type converters. If they can make them durable then I think they will ultimately be the way to go, and perhaps those will be an enabling technology for larger cells. With their history of poor reliability, though, I think microinverters will require several years of solid performance in the field before they will become widely accepted by installers. (In my one experience with microinverters we lost our first inverter within about a month and all of them, in a group of 20, in less than a year.)--Squirmymcphee (talk) 13:17, 31 May 2009 (UTC)Reply
I just watched this video by Enphase [8]. The installation process seemed painfully slow. Stickers? My god man. I've always assumed these micro-inverters would be built-in at the panel factory or at least be engineered to be removable cartridges. Bad inverter? Slap in another one. Done. This seems obvious to me. Mrshaba (talk) 00:01, 31 May 2009 (UTC)Reply
I agree with you that that's the way it should be, and I'm guessing Enphase would agree with you. But at the moment they're probably finding it easier to sell inverters to end users than to module manufacturers. If I'm a module manufacturer then I want to see how the Enphase inverters perform in the field before I even think about integrating them into my product. That said, competition in the PV industry now is such that some high-cost module manufacturer looking to differentiate itself might latch on with somebody like Enphase and take a flyer on it. That wouldn't surprise me in the least, though it might make or break both the module manufacturer and Enphase.--Squirmymcphee (talk) 13:17, 31 May 2009 (UTC)Reply

Does a 4x4 silicon solar cell produce the same voltage as a 8x8 cell or a 12x12 cell? About .5 volts? Can you divide a cell using a chemical cut i.e. use an insulating spray that would penetrate into the wafer and electrically isolate one region from another so that you produce 2 or 4 or however many electrically distinct regions within a cell without having mechanical separation. Mrshaba (talk) 22:07, 1 June 2009 (UTC)Reply

The voltage is determined by the number of junctions, so yes, a silicon solar cell produces about 0.5 volts regardless of its size. If you cut through the p-n junction you have the functional equivalent of two solar cells. But then you're stuck with the problem of connecting the + terminal of one of them to the - terminal of its neighbor (i.e., running a tab from the front of one to the back of the next). It's much easer to do that when the cells are physically separated, but actually cutting all the way through them is both slow (even with a laser) and a yield risk. Then you either have to integrate them into modules yourself or find a module manufacturer willing to buy an unusually shaped product. None of these issues is insurmountable, of course, but the solutions all cut into the savings you achieve by going to a larger cell in the first place.--Squirmymcphee (talk) 19:49, 2 June 2009 (UTC)Reply
Thank you for the explanation. So, could you use an acid treatment or laser to cut through the top layer (20 micrometers?) of a cell to achieve electrical isolation between zones within a single wafer? Could you explain what you mean by "unusually shaped product"? Mrshaba (talk) 07:21, 3 June 2009 (UTC)Reply
Hmmm...Something like this perhaps [9]. Mrshaba (talk) 07:55, 3 June 2009 (UTC)Reply
You can certainly cut through the top layer to achieve electrical isolation -- in fact, it is routinely done in PV production today to isolate the back surface from the front (the n+ diffusion wraps around to the back in most cells and needs to be interrupted to prevent shunting). The difference between that and isolating several solar cells within a single wafer, though, is that in this case isolation is performed as close to the edge as possible to avoid creating multiple cells in each wafer. I had to think about what I meant by "unusually shaped product" -- I guess I meant that the resulting cells would have unusual grid orientations and/or tabbing requirements that could pose problems for traditional tabbing and stringing equipment.
There have been a few variations on the v-groove solar cell over the years (see also vertical junction cell and sliver cell). The sliver cell has come the closest to being commercially viable, primarily because it relies on silicon and not on III-V materials and expensive processes. However, I don't think any of them is faring too well from a commercial standpoint.--Squirmymcphee (talk) 20:57, 10 June 2009 (UTC)Reply
edit

Thank you for sorting that out. Interesting info as always.

I was digging around some old earnings statements and found that ScanWafer was selling wafers for $.92/Watt back in 2001/2002 (assuming my currency conversions are correct). Times have changed eh? As the cost of wafers has gone up we've seen manufacturers attack the wafer thickness issue but now that wafer prices are diving I'm wondering what new approaches manufacturers might take to lower costs? I'm sure engineers will continue to attack the thickness issue but probably not as much. What will the unexpected angles be? You pointed out that UMG will be ditched. That's making more sense to me. What else? Might mono-crystalline module makers square off cells to increase packing density? My non-scientific photoshop test indicates you could get 2-3% more cell surface area in a module with squares vs. pseudo-squares. How close is that approach to making sense? What other angles? I still like the idea of increasing the size of wafers and I did find an interesting tidbit on this. You mentioned that Sharp had experimented with larger cells. I wasn't able to find any documentation of Sharp's work but I did find some info on Q-cells working with 8 inch cells. Here's the quote, "*Due to the combination of higher cell thickness and the current silicon shortage, we do not expect a significant revenue share for 8-inch-cells before 2008." This quote seems to indicate that there's a cross-over point at which the reduction in processing costs off-sets the extra silicon costs associated with thicker wafers. There's still the pesky voltage issue but perhaps bigger wafers will become fashionable someday. Mrshaba (talk) 07:20, 11 June 2009 (UTC)Reply

One other random question. Have you ever heard of a CZ furnace that spins the crucible instead of the ingot? Mrshaba (talk) 08:44, 11 June 2009 (UTC)Reply

I don't think mono manufacturers will start squaring off wafers. Silicon is getting cheaper, but it isn't getting that cheap and besides, the resulting increase in module efficiency isn't worth too much more than a press release. The drive toward thinner wafers is stalling, though, while manufacturers work on getting some of the cruft out of their processes. And I think that's where you will see a lot of work in the next year or so -- not the stuff people notice, but the stuff that has been ignored by the manufacturers have they have sprinted headlong into building as much capacity as possible as fast as possible. Right now there are a lot of manufacturers who do what they do for no other reason than because it works, but now with cash tight and margins shrinking there is a lot of pressure to optimize. The low-cost manufacturers will look for the next thing to maintain their advantage -- Suntech with Pluto cells, for example -- while those who are a bit behind will try to squeeze as much performance as possible out of their existing lines without spending more money (e.g., by realizing they use too much silver in their busbars, or that they can diffuse less phosphorus and get the same -- or better -- results). I know it isn't sexy, but with the focus on expanding capacity for the past 5 years or so there is a lot of fat to trim. Finally, in new production I think you will see a renewed focus on commercializing more advanced cell designs instead of just getting conventional product into the pipeline as fast as possible. Look in particular for cells with electroplated contacts, with dielectric passivation on the rear, and for growth in passivating layers other than silicon nitride (e.g., amorphous silicon, aluminum oxide).
And no, I haven't heard of a Cz furnace that spins the crucible. I'm no metallurgist, but I would think that would totally change the dynamics of the crystal growth. That's not to say it couldn't be done, but it isn't really my area.--Squirmymcphee (talk) 18:09, 12 June 2009 (UTC)Reply
I asked Ted Ciszek the spinning crucible question. He said the standard procedure is to spin the crucible opposite the ingot. That pretty much answers my question. I asked him if there's a spinning crucible only process but haven't heard back yet.
I appreciate your insight on where things are going. Your view on PV is way way way beyond what I can see or imagine. I work in transmission and distribution in British Columbia... We are many years away from widespread residential/commercial PV up here but I still enjoy thinking about things. Even in the Great White North we see PV gadgets popping up here and there. The family went to the beach yesterday for fish and chips and paid for parking at a solar powered meter. With module prices diving I figure we'll see many more day to day solar powered gadgets. We'll probably get an acronym for them. PVGs? SPG? Something like that... After working in numerous acronym driven industries for many years I'm getting rather tired of lame acronyms but I just sigh and nod and sigh... Call it what you will..
I'm really not convinced on the square mono-wafer thing. It seems cost optimization would need to compare ingot growth against upstream costs. Obviously, if the extra ingot growth costs are more than the upstream wafer/cell/module savings you go with semi-square wafers. When I look at LDK and ReneSola's processing costs it seems like square mono-wafers aren't "a press release". With processing costs approaching $.25/Watt square wafers look like the way to go. Just an outsider's thought. Mrshaba (talk) 05:25, 15 June 2009 (UTC)Reply
Opposite the ingot ... that makes sense -- I know you need some sort of tangential flow of the melt around the seed and I cannot imagine you get a very strong one if you're only rotating the crucible (unless it is a very small crucible), since the crucible walls are so far away from the seed. Which reminds me of a trend I neglected to mention: large ingot sizes. Not so long ago a 150 kg ingot was considered large. Now 240 kg is pretty standard, and some equipment manufacturers are claiming that a 600 kg ingot gives a 50+% reduction in ingot-casting costs.
A full mono square will throw away nearly 37% of the wafer, and you need a 12-inch round wafer to produce an 8-inch square. Whatever savings you get upstream not only have to overcome the additional cost of the discarded/recycled/remelted/whatever silicon, but you have to compare it to what your costs would be if you used larger wafers and processed fewer pieces to achieve the same annual MW production. I understand what you're saying, and I've heard people inside the industry say the same thing -- but then I've also heard people say exactly the opposite (and in fact I believe there are a few small manufacturers that refuse to throw any silicon away and use completely round wafers in their modules). I think the majority of the mono industry will stick with pseudo-squares for the forseeable future. They dominated back when silicon was less than $20/kg, and while I suppose the thinner wafers in use now might change the math a bit I see no reason to think they won't continue to be the most popular choice.--Squirmymcphee (talk) 20:58, 17 June 2009 (UTC)Reply
Hmmm... Bigger ingots reduce casting costs - makes sense to me. Mind you, a reduction in ingot costs would support (incrementally) a transition to square wafers. My simple math on square vs. semi-square wafers had square wafers "wasting" an additional 20% of the wafer - I'll have to recheck that math. Perhaps I'm missing something else? Aren't the sawed off semi-hemispheres recycled back into ingots? Is this not the case? Are they truly wasted?
Here's a quote from, The Handbook of Photovoltaic Science and Engineering (2002). "The lowest (publicly offered) module selling prices in 2002 were about $3/WP... The wafer itself represents about 65% of the module cost, approximately equally divided between purification, crystallization, and sawing."
Now that I look at the quote again I can't reconcile it with what I've read in the earnings reports from ScanWafer which listed wafers ASPs of 92 cents/Watt in the same time frame? I guess I'm going to have to double back and reread the fine print in the ScanWafer reports as well. Thing is, Rogol also listed wafer ASPs at $1/Watt around 2003ish. Now I'm really confused. Ugh...
My overall logic is that ingot costs "look" like they're currently in the neighborhood of 17 cents/watt. If you had to make 20% more ingots you'd spend 3.4 cents/Watt more in ingot costs but I think these extra ingot costs could be recovered in the wafering/cell/module stage. The numbers come out mighty close but they are based on multiple assumptions. I'll double check the sources and ask LDK/ReneSola to clarify their processing cost estimates. Mrshaba (talk) 01:23, 18 June 2009 (UTC)Reply
On the topic of big ingots here's a quick and timely mention of LDK's 660 kg ingots. Do you have any sources that break down ingot costs/Watt? i.e. Where did you hear that 600 kg ingots result in a 50%+ reduction in cost? I'll look around.[10]
Correction: I recalculated the ScanWafer 2002 ASPs. I came up with $1.02/Watt which makes more sense than $.92/Watt.[11]
Do you have any idea how the ingot costs compare between multi and mono ingots? I asked ReneSola about this specifically - crossing fingers.
Is "casting" a term reserved for multi ingots or can it refer to both multi and mono ingots? Mrshaba (talk) 02:35, 19 June 2009 (UTC)Reply
Trina's Q109 Conference call mentions that mono has $.10/Watt higher costs than multi ($.20/Watt higher when depreciation is considered). I had reasoned that mono modules have less area per watt so less glass, aluminum, silver etc. This is probably true but perhaps the processing steps are that much more expensive. If mono already has higher non-silicon costs (which I assume to mean everything past wafering) there would be no reason to square wafers. Mrshaba (talk) 20:32, 21 June 2009 (UTC)Reply
It's a preliminary report but LDK's wafer ASPs have reached $1/watt - perhaps as low as $.93/Watt. [12] That's down from $1.54/Watt in the first quarter. Mrshaba (talk) 00:31, 6 July 2009 (UTC)Reply

Can't recall precisely where I read about 660 kg ingots cutting costs by 50% -- it was probably an article in a trade magazine or a press release. I see a lot of those things and have been a bit busy at work lately, so it's hard to keep track. To me, the quote from the Handbook of Photovoltaic Science and Engineering sounds more consistent with Cz than with multi -- at one time, anyway, wafering was the biggest cost with multi. Of course, most of the published information comes from a day when wafers were thicker, so those sawing costs are now spread over more wafers and it may now be a different story. I'm a little out of touch on that aspect these days.

I want to say mono ingots are some 20-40% more expensive than multi ingots, but I wouldn't quote me on that. Mono wafers tend to produce higher efficiencies, of course, but silicon utilitization tends to be higher with mono ingots as well (despite the pseudo squares), so the cost per watt works out in the end. As for terminology, strictly speaking mono "ingots" are boules that are grown, not cast. Multi, of course, comes in ingots that are cast. In practice, though, the terms get tossed about pretty loosely.

Part of the reason mono is more expensive than multi is because it produces higher efficiencies, and manufacturers find that they can charge more per watt for high efficiency cells than for low efficiency cells. Therefore, if they are using mono it is likely that they find it worthwhile to spend a few extra cents/watt to bump the efficiency up. Why not do the same with multi? Because getting the same efficiency bump would cost more and bring less revenue. It seems a bit paradoxical, but as an example the folks at BP Solar told me long ago that buried contacts would work fine on multi but weren't worth it economically, which is why they always did them on mono (except, I think, for a short period, at least in R&D).

As for wafer prices, I know at least one company anticipating $0.80/Wp mono by the end of the year. Only those with new contracts and those who successfully renegotiate their existing contracts will get those prices though -- some will be stuck with high prices for a long time.--Squirmymcphee (talk) 20:16, 13 July 2009 (UTC)Reply

Hey there Strangermchpee... busy with work eh? I hear you. I've been busy myself and haven't had a spare day in several weeks. The $.80/Wp figure sounds about right but who knows these days. I'm patiently awaiting the next round of earnings statements but I'm not expecting much of any good news. My gut tells me we'll see some big bancrupcies/restructuring/etc. by year's end. There's only so many times you can go to the well for additional financing. When I ask myself, are we in the calm before the storm or the calm after I decide more and more on the former. I guess it's this negative vibe I have that's making me think poly/wafers prices will go lower than our best guesses. There's probably a cliche out there that goes something like: hunger makes you hunt better. And so.. despite my negativity, I think these lean times will ultimately be good for the industry. Mrshaba (talk) 04:04, 14 July 2009 (UTC)Reply
Definitely agree that these times will ultimately be good for the industry. I have been waiting for these times for years now, and the fact that they had yet to come along are a major reason I have avoided taking a job with a startup. My boss recently complained that he's fielding a couple of calls every day from startups claiming to have the solutions to all of our problems, if only we'll license their technology -- that says to me that a lot of these startups will be finished soon. For us, these times have meant we can take a breather from trying to ramp production as fast as possible and focus on cutting costs and improving efficiency. By the end of 2009 we will be making a significantly better and cheaper solar cell than we were at the end of 2008, and will be completely a result of tweaking our processes: changing a temperature here, saving a microgram there, that sort of thing. It is amazing how far fundamentals will get you when you have the time to apply them.--Squirmymcphee (talk) 20:06, 16 July 2009 (UTC)Reply
 
Maybe not.
  
 
No Comment
I was reading this article today and noticed the non-traditional packing arrangement in the pictured module. [13] It got me daydreaming about the packing density of pseudo-squares and I realized you could improve your packing density within the core of your module by mating the "corners" of the squares. This results in a staggering effect that would hurt you at edges but... aha... you could account for part of the staggering by having romboid shaped modules - I realize it's a funky shape but rhomboids tessellate and some architects/customers might prefer the shape. Maybe...
Then I thought why not use two different sized cells to create psuedo-psuedo-squares. You'd get some improvement in packing density but then I wouldn't expect two separately sized wafering lines to operate more efficiently... It was just a thought...
Here's a question for you. Why not use a transparent substrate and a recessed mirror behind the cells to reflect the light that gets through between the cells? Has anyone tried this type setup? Mrshaba (talk)
One other thing I wanted to ask you. Does squaring/psuedo-squaring of wafers cause stress effects at the corners of the cell that significantly degrades performance? Also, why don't we see any hexagonal cells? Is this shape difficult to work with? Mrshaba (talk) 01:21, 31 July 2009 (UTC)Reply
Interesting ideas. I hate to always be a downer, but I see big problems with all of them. Staggering the cells would seriously complicate the stringing process, which is already far more complicated than anybody likes. This could be addressed by using rhomboid shapes, but that implies rhomboid-shaped cells and I shudder to think at the negative effect that would have on yields. The acutely angled corners would not fit well in wafer carriers and would most likely be a nightmare with certain types of automation, and those corners would almost certainly tend to break a lot. The two differently-sized cells idea is tough in that the two different sizes would have to lie in separate strings to avoid serious mismatch losses because of the different amounts of current produced by each cell -- all of that with the constraint that the number of cells in any given string, regardless of cell size, be equal so they could be connected in parallel in the end. Plus, you'd be producing smaller cells at a higher cost per watt just to take up a few more percent of open space in the module. I feel pretty confident in saying that in terms of cost it would be a net loser.
You say "transparent substrate" and "between the cells" -- presumably you're talking about ordinary silicon solar cells in a module that is transparent in the back, with the mirror reflecting light back into the module. This has been tried, more or less -- the white backing materials that are used in some modules are quite reflective and have been shown to improve module performance by scattering light back into the cells. It is most effective with some sort of textured glass on the front of the modules to aid with light trapping, much like the textured surfaces of solar cells, but textured glass has problems when it comes to field deployment.
Squaring/pseudo-squaring generally does not produce any significant stress effects in most substrates. The actual process of cutting the silicon can introduce microcracks or interfere with internal stresses in a way that can cause wafers to break later, and surely in theory these stresses can interfere with performance. In practice I am not aware of anything indicating that it is a major issue, and certainly nobody I know worries much about it (except from a breakage standpoint).
As for hexagonal cells, I believe a pseudo-square is less wasteful of silicon and requires fewer cuts. It may or may not pack into a module better -- I haven't done the math. I actually have had occasion to work with hexagonal cells when I was in school and my wafers wouldn't quite fit into the furnaces, so I diced the corners off. It was a major pain, though I wasn't using the sort of process that would be used in a production line.
One other problem that anybody with a design that affects stringing is going to face, no matter the reason or the upside: Module manufacturers want to be sure they can get cells that are compatible with their tabbing and stringing equipment, and equipment manufacturers want to make sure they have customers for their machines. Therefore, module manufacturers are often reluctant to buy cells that don't fit standard tabber/stringer dimensions. A lot of PV companies with unusual designs, no matter how good -- back-contact cell companies come to mind -- have trouble selling their cells because they have unusual tabbing and stringing requirements. They could make the modules themselves, but to do that they might also be stuck designing and building the processing equipment themselves. It takes a vertically integrated company to unilaterally move forward with an unusual contact design.--Squirmymcphee (talk) 18:51, 4 August 2009 (UTC)Reply

Not a downer at all. Those packing arrangements were geometrical daydreams. I figured the processing problems would far outweigh any packing density advantage but I felt like sharing... The hexagonal cells I wasn't sure about but I'll take your word on the tabbing issues they would raise. Aside from the these packing density questions I well realize that when you get down to it you really need to think about the "global efficiency" of the manufacturing process above all else.

I finished reading The Selfish Gene today... Great read. Dawkins talks about the evolvability of species being a primary overall indicator of fitness. I've thought of this evolvability aspect before in regards to PV but never used this word to describe it. I'll have to think about it more but it seems to me you could use much of the same language and reasoning on PV vs. X-energy technology that Dawkins applies to genes vs. alleles. One other thing... Long ago I asked you if you knew of a database that listed if a given scientific paper was cited by subsequent papers - Dawkins mentions just such a database called the Science Citation Index. Mrshaba (talk) 03:38, 5 August 2009 (UTC)Reply

"It is most effective with some sort of textured glass on the front of the modules to aid with light trapping, much like the textured surfaces of solar cells, but textured glass has problems when it comes to field deployment."

I was thinking the backside mirror would be the textured part. Mrshaba (talk) 03:41, 5 August 2009 (UTC)Reply

::Do you happen to know which purification route is cheaper to get to solar grade silicon - Siemens or FBR? Mrshaba (talk) 01:11, 31 August 2009 (UTC)Reply

Any thoughts on Suntech's 16.53 efficient m-Si modules. I'm impressed I must say. They say their Pluto process requires more automation - more than they expected even. I've been wondering if the extra automation is what's delivering the efficiency gains in addition to process itself. I've always expected automation to deliver a cheaper product eventually but I never thought it would deliver a more efficient product. What I mean is I never tied the two (automation and efficiency) together but I guess it makes sense. What's your opinion on that? How much pressure do you think there is to automate and how fast will PV fabs become mostly automated? Which stage first, second third? I've heard that wafers have gotten so thin it's becoming difficult to handle them manually. I can't recall who it is but they're developing air handling systems that use Bernoulli suction to move the wafers around - that took me right back to an exhibit at the San Francisco Planetarium... Neat!!! Mrshaba (talk) 02:40, 24 September 2009 (UTC)Reply
I suspect the additional automation goes into creating the selective emitter (i.e., the extra-heavily doped regions under the front contacts) and aligning the contacts to the heavily doped regions. Not that getting the alignment is what requires the automation -- it is supposedly a self-aligning process -- but they need to add some process steps to get it all done. In other words, I would read "additional automation" as "extra process steps." In a modern PV fab, if they need an extra wet bench, for example, it's going to be an automated wet bench. PV fabs are already quite automated, at least once you're above about 10-25 MW in size, with the possible exception of module assembly. That, without a doubt, is still where most of the manual labor exists. Where I work, for example, one of our cell fab lines has a 60 MW capacity and is run by a crew of just 8 operators, a shift supervisor, and an assistant. If it were a bigger line we could have more automation and fewer operators per MW. As for the thinness of the wafers, when I first stepped into a solar cell R&D lab, 350 um was standard and everybody was terrified to handle 250 um wafers. Nowadays I routinely handle 160-180 um wafers with little fear (or breakage) and 350 um seems laughably thick to me. A lot of it is just a matter of getting used to it. Of course, there are challenges for automation systems, and those challenges always change with the thickness. At current thicknesses, separating wafers with automated systems is always a problem, and various air- and vacuum-driven systems (including Bernoulli systems) are common ways around the problem. Bowing after firing of the metal contacts is also a serious issue -- really thin wafers, because they are so flexible, can really curl up like a potato chip if you aren't careful.--Squirmymcphee (talk) 16:17, 17 October 2009 (UTC)Reply

I'd like to clarify something. You described the process of "creating the selective emitter" as "supposedly self-aligning". Pluto cells are glorified LGBC cells right. If you've got grooves in your cell you more-or-less automatically have alignment for the selective emitter diffusion step and contact formation? I'm confused that you'd say "supposedly self-aligning". Am I missing something? Mrshaba (talk) 06:19, 22 December 2009 (UTC)Reply

Modules

edit

Thanks for the detailed info. Can you estimate the manpower requirements at the module assembly stage? Why does module assembly require more manpower? I've asked you about larger modules before but you mentioned they are unwieldy. Do you think this unwieldiness is more of a problem during manufacturing or installation? To put it another way, do you see rooftop systems using larger modules or have we pretty much reached the module size limit? Mrshaba (talk) 02:03, 18 October 2009 (UTC)Reply

Let me clarify on the unwieldiness point: I think there is a practical size limit, but I think it is much larger for multi-megawatt industrial installations than for small residential installations. In general, I would say that large size is a bigger problem during installation than during manufacture (though gigantic pieces of glass become quite expensive per unit area), but when you are sending many truckloads of modules to one site a larger module cuts down on overall installation costs. On the other hand, that advantage becomes a liability when you need an oversized vehicle to transport a handful of modules to a private home. In the residential case, you're also more frequently dealing with roof protrusions and the like, and using very large modules can, in some cases, effectively reduce the useable roof space. All that said, I'm sure you're aware of companies like Applied Materials pushing their very large modules recently. It is tough to say where the limit is.
As for the manpower requirements, that is tough to estimate. It depends in part on where the module is being manufactured, for one thing -- where labor is cheap there is less incentive to use automation. Throughput requirements are also much lower, by a factor of about 30-75, depending on the number of cells per module, so in many factories module assembly may still be a bit below the line where automation provides significant economies of scale. And for both of those reasons, plus the fact that modules are larger and in some ways more complex than individual cells, module assembly is where automation still has the longest way to go. So a lot of the handling that is done automatically at other steps in the process is done manually at the module assembly stage.--Squirmymcphee (talk) 18:28, 10 November 2009 (UTC)Reply
Hello SquirmyMcphee... How's the Fall treating you? It's cold and wet up here - brrr... And dark coming and going from work - grrr...
Contractors love their super-sized trucks so I'm not convinced larger modules will significantly add to shipping costs/difficulty - within reason of course. I figure ergonomic considerations will limit module sizes before shipping becomes a concern. Roughly 90% of the market is 38" - 40" by 58" - 64" and 45-50 pounds. By contrast, half inch plywood sheets are comparable in weight but are 55% larger at 48" x 72". From an ergonomic perspective I can see modules getting larger but only if weight came down proportionately. Just my thoughts.
I got hot and bothered over roof protrusions a few weeks back - I had alot of free time on my hands. If average system sizes continue their upward trend (roughly 5.2 kW at the moment) we'll run out per-unit south facing roof space. I wrote to one fellow who told me average residential system sizes will probably plateau at about 10 kW. How big of an issue do roof protrusion become as system sizes plateau - a kW maybe? Overall it's the kind of problem you want to have. The industry can cross this bridge with the building code people when they come to it.
Back to the manpower issue. I'm looking for an approximation. Put it this way, let's say you've got a state of the art wafer-to-cell/cell-to-module PV facility. How would 100 workers proportion out between the cell and module stages? Where do you see this proportion is 5 years? Mrshaba (talk) 22:01, 13 November 2009 (UTC)Reply
A 25 MW wafer-to-cell line can be run with 7-8 operators/shift (that is, not including supervisors and maintenance techs). To my knowledge, a similarly sized module fab needs more like 10 operators/shift. On that basis, your 100 workers would be split 45/55 in cell/module production. However, by the time you reach scales where you are running 100-worker shifts I think you will see more automation on the module end of things. Since there are fewer discrete production steps in the module assembly process, that might bring the labor requirement down substantially. There are plants that are operating on such scales, but I have not had the opportunity to visit them and I do not know what they do at the moment. All that said, the throughput of cell production equipment is rising as well, so I don't think you will see a massive shift in the proportion -- maybe it will flip to 55/45 cell/module. Pure speculation, that.--Squirmymcphee (talk) 16:59, 6 December 2009 (UTC)Reply
Thank you for the speculation. I didn't know the manpower requirements were so close. I've been wondering why Suntech set up a module fab vs. a cell fab in Arizona. Now this makes more sense to me. Mrshaba (talk) 20:27, 6 December 2009 (UTC)Reply
With lower labor prices in China, I think Suntech can probably afford to use more manual labor in both cell and module manufacturing in plants located in China. The Arizona fab was done as much for political and marketing reasons as anything else, I think -- they want to be able to participate in US-government-subsidized projects and are getting worried about the outcry over such money going overseas. And I imagine that plant will have more automation than an equivalent plant in China.--Squirmymcphee (talk) 21:20, 13 December 2009 (UTC)Reply

I noticed that ECN/REC set a new m-Si module efficiency record of 17%. [14] I'm impressed by how much the efficiency jumped the last two times this record has been broken. A few months back it jumped by about 1% and this time it's jumping by .5%. This seem like a lot. Do you know why there's so much action in this area all the sudden? Is this just coincidence or is there more to come?

I looked around for the theoretical efficiency limit for multicrystalline silicon but couldn't find a value. Do you happen to know that number off hand? Can you offer a guess where the practical efficiency limit will be? Mrshaba (talk) 10:00, 12 December 2009 (UTC)Reply

A few research groups looked at the big gap between cell efficiency and module efficiency and decided there was a research opportunity there. Manufacturers have always thought in terms of module power rather than module efficiency, so little attention had been paid to aspects of module design and construction that affect the efficiency figure (as opposed to the power figure) and there was plenty of low-hanging fruit. There is a marketing advantage in being able to claim a high module efficiency, but the technicial and economic advantages are there only to the extent that power losses can be eliminated and the size of the module can be substantially reduced. ECN makes a big deal of some of these losses, but in reality not every efficiency loss is a power loss when you go from cell to module.
As for the theoretical efficiency limit for multicrystalline silicon, it is exactly the same as that for monocrystalline silicon -- roughly 32% (depending on whose accounting you follow). After all, multicrystalline silicon is simply made of chunks of monocrystalline silicon that are packed together. What makes multicrystalline silicon inherently lower in efficiency than mono is greater impurity concentration and recombination at the grain boundaries. However, when computing the theoretical maximum you assume zero recombination anywhere, so these differences are wiped out. I feel confident that 21% is possible for multi in the lab, and I think somebody will probably crack 18% for an average production efficiency on multi sometime in the next 3-5 years.--Squirmymcphee (talk) 21:20, 13 December 2009 (UTC)Reply
Thanks for the insight. Could you explain what you mean by: not every efficiency loss is a power loss?
Here's something funny I spotted. For a while I've wondered where Suntech came up with the name Pluto for their high-efficiency cell technology. Apparently the name is a play on BP's Saturn cells. But that's not all... In Chinese, Japanese and Korean Pluto translates as underworld king star (冥王星) Some Indian languages use a form of Yama, the Guardian of Hell in Hindu mythology, such as the Gujarati Yamdev. Nyuck nyuck nyuck Mrshaba (talk) 10:22, 14 December 2009 (UTC)Reply
Sure. Suppose you have 72 solar cells and you build two modules of 36 cells each. In Module A, you leave 2 mm between each cell; in Module B, 4 mm. The modules are otherwise identical. They will both produce the same amount of power, but Module A, since it is slightly smaller, will have a higher efficiency.
Of course, you could make a case that the light lost in the larger gaps between cells is a power loss, and ECN does make that case, but from an end-user perspective I'm not sure it matters much. In most cases end-users just want to get the amount of power they need, and if roof space is at a premium they will select modules with high-efficiency cells in them. A couple of centimeters per module is not going to be a make-or-break factor, IMO.
Interesting bit on the origins of the Pluto name. I did not know that.--Squirmymcphee (talk) 14:25, 29 December 2009 (UTC)Reply
Your explanation of the power vs. energy thing makes perfect sense.
I found out some additional details on the Saturn vs. Pluto relationship. BP's Saturn cells were LGBC cells and used technology licensed from NSW. Pretty much the same goes for Pluto cells. So, the Pluto name implies more than a simple play on words. I still think it's funny that Pluto translates as underworld king star though.
I posted this question up higher but you might have missed it. You described the process of "creating the selective emitter" as "supposedly self-aligning". Pluto cells are glorified LGBC cells right. If you've got grooves in your cell you more-or-less automatically have alignment for the selective emitter diffusion step and contact formation? I'm confused that you'd say "supposedly self-aligning". Am I missing something? Mrshaba (talk) 22:54, 29 December 2009 (UTC)Reply
Pluto cells are actually based on PERL cells, not LGBC cells. The selective emitter is formed not by diffusion but by a laser doping process -- Suntech seems to be the first to have gotten lasers with the required throughput. The laser doping occurs after antireflection coating deposition; then, the laser is used to simultaneously remove the antireflection coating where the contact lines will be and melt the silicon underneath. The molten silicon is not ablated, but allowed to recrystallize, so little to no grooving occurs. The doping level in the recrystallized silicon increases -- they don't say how, but it is probably either by activating inactive dopant left behind by the initial emitter diffusion or by adding additional dopant on top of the antireflection coating prior to the laser step. Then they electroplate contacts onto the exposed silicon. The process is self-aligning because the metal will only plate on the exposed silicon, not on the antireflection coating. That, at least, is my understanding of the process. They have been secretive enough about it that I doubt many people outside of Suntech know the real story.--Squirmymcphee (talk) 23:07, 2 January 2010 (UTC)Reply
My apologies, I thought Suntech's version of PERL incorporated laser grooves as well. I know Suntech's press releases generally mention the PERL process but they also go on and on about the texturing and the extra-fine "fingers" - I thought these "fingers" were laser grooves and the vertical channeling was the reason Suntech could make thinner grid lines and use copper. I must have read too many of Finlay Colville's laser stories. Sorry about that... Just when I thought I was getting the hang of things too. I guess the Saturn/Pluto thing threw me off.
So... Is there any reason for Suntech's PERL type cell to bury its contacts or have they somehow managed to get thin grid lines without having to use vertical channels? What I mean is - neglecting costs, do you think laser grooves would improve Pluto`s efficiency even more.Mrshaba (talk) 01:37, 3 January 2010 (UTC)Reply
No, I don't think laser grooves would help; in fact, I think they'd probably hurt. The primary reason to use them in LGBG cells was to effectively turn the grid fingers sideways and reduce the amount of area coverage on the front of the cell. The Pluto design uses plated contacts that are just as narrow as the LGBG contacts, but without having to cut the groove. The problem with cutting the groove is that it is fairly deep and, therefore, requires an IR laser. These create a good bit of damage that can penetrate quite a few microns deep into the cell. The Pluto cell gets essentially the same effect, but without the damage.
Of course, another way to to look at it is this: If cutting grooves would improve the efficiency, UNSW would have done that with the original PERL cell years ago and pushed its record efficiency even higher.--Squirmymcphee (talk) 12:31, 17 January 2010 (UTC)Reply

You should write a book. I'm not pulling your leg either. You are a great writer and an extremely knowledgeable professional. PV needs a flagship read. I know of no one who can explain this subject from top to bottom as well as you. You'd make a mint. Good luck if you're working on something. Mrshaba (talk) 20:48, 26 January 2010 (UTC)Reply

Thanks for the compliment! I actually would like to write a book, and I have several ideas and plenty of text fragments, but so far nothing coherent enough to call a manuscript or even an outline. My ideas have ranged from technical books that would primarily serve a limited professional and academic audience to things meant for a much wider audience. I'm curious: What sort of thing would you consider a "flagship read"?--Squirmymcphee (talk) 21:06, 11 February 2010 (UTC)Reply
What's a flagship read? Hmmm... Some thoughts... Two books that could provide some structural guidance are The Prize and Cadillac Desert. Both great reads with a solid cast of characters which I think you'd also need in a great book about PV - engineers/scientists, entrepreneurs and perhaps a sympathetic government official or two. Who fits this bill for PV? The Bell Labs team, Green, Zhengrong, and Sheer come to mind immediately but I figure you can think of many more. If I was sketching out a story line I would think up a cast of characters that could be sprinkled strategically throughout. I'd pick a few quirky oddballs where possible.
I'd expect the book to be comprehensive. Everything from pn junctions to solar farms would be mentioned. To do that you'd need to break everything down into fundamental bits that you could rebuild anyway you wanted. The most wonderful thing about PV for me is how the technology is evolving so I think you'd need to draw a solid picture of PV's trajectory.
I've read several books and articles on PV that meander through the history of power plants, peak oil, the Smart Grid, Destertec type schemes and global warming. I think all of these avenues are distractions and should be avoided for the most part. I'd rather read a book about PV that didn't rely on any crutches. No apologies - no excuses.
Just some thoughts. Hopefully some of this makes sense. Mrshaba (talk) 23:10, 13 February 2010 (UTC)Reply
Thanks for the thoughts. I'm not familiar with the books you mention, so maybe I will check them out. The sort of comprehensive story that you talk about is a pretty tall order when you consider how fast the industry is moving, but I think it is achievable if the story is rooted in fundamentals. I mostly agree about limiting the scope purely to PV, though I do think it is important to motivate the need for PV with some context. However, I think that can be done succinctly without going off on any great tangents. Something I will mull over....--Squirmymcphee (talk) 20:41, 16 February 2010 (UTC)Reply
The Prize earned a Pulitzer and Cadillac Desert is right up there. I mention the two because both books are flagship reads; The Prize is about oil and Cadillac Desert is about hydro. A PV flagship (as I imagine it) wouldn't have as much history but it would still need great characters to move the story along. Real people resonate with the audience.
I agree that one of the toughest parts would be contending with the pace of this subject but I think you've got the right skill set to tackle the problem. Mrshaba (talk) 15:43, 18 February 2010 (UTC)Reply


Energy Payback

edit

Have you seen anything lately that examines the energy payback of PV? The most recent studies I can find go back to 2005. Mrshaba (talk) 16:06, 1 March 2010 (UTC)Reply

Do you think there will be any spillover from LED manufacturing into PV? Mrshaba (talk) 17:17, 1 March 2010 (UTC)Reply
Seems like Erik Aselma from the University of Utrecht is always publishing something on energy payback. I haven't been tracking it, though, so I can't suggest anything offhand. Sorry. As for LEDs, it seems possible that some thin-film deposition techniques could spillover, but it's tough to say -- throughput would be the issue, I think.--Squirmymcphee (talk) 17:15, 11 March 2010 (UTC)Reply

Caltech

edit

Caltech isn't the University of Utah, so I expect we're not having a reprise of that. At least I hope so.

I've reread the press release and still get from it what I read at first.

I understand that you're making a projection from the theory (except for the hearsay about it you can't quite recall) but ... I'm so far from physics (and never took any solid state physics at all) so will not argue the technical side of things.

If what you think is what they've got is so, then why bother with the hooha? it's an improvement and maybe a nice one, but then lots of folks, including two in Israel are claiming PV improvements as well.

As I get it, they're not even talking about absorption via the standard mechanism Einstein got his Prize for, but something else having something to do with the exact spacing between the silicon nanorods. Some field effect? Something like GMR or something new like that? Is there a Nobel here?

Can I be so bold as to suggest a phone call, physicist to physicist? And a report here on what the press release is really talking about? ww (talk) 03:49, 8 March 2010 (UTC)Reply

No, Caltech is not the University of Utah, and this is most certainly not a reprise of that. As I said before, the press release is poorly worded. It refers to 85% (or whatever) collection efficiency, which is a very different thing from the power conversion efficiency. But since you weren't satisfied with my earlier explanation I had a look at the Nature Materials paper referred to in the press release. You can see in the publicly available abstract that they claim only that their design "may offer increased photovoltaic efficiency". In the paper itself, the maximum value they put on this efficiency is 17%.
The hooha is mainly for three reasons: (1) they can claim higher quantum efficiency and collection efficiency than similar designs have achieved in the past, which is important from an academic standpoint; (2) Caltech needs to give its supporters a reason to donate money; and (3) designs like this, which use very little silicon while producing high efficiencies, have tremendous potential if they can be successfully commercialized. Why would this improvement get a big press release and the Israeli improvements not? I don't know, but I can say that universities and research institutes tend to issue press releases over anything that gets published in one of the Nature journals, Science, and a handful of others, as they are very prestigious journals and it looks good to donors.
As for the absorption mechanism, when you get right down to it, it honestly is no different from the standard mechanism that all crystalline silicon solar cells use. However, you're correct that there is something special about the geometry of the nanorods -- they form what is called a photonic crystal that is capable of trapping light in very thin layers of material, where it just bounces around until it is finally absorbed by the silicon.
If anybody gets a Nobel for this sort of thing, it will be the guys who invented the modern photonic crystal 20+ years ago (though photonic crystals have been known for more than 100 years). The Caltech design is not even the first solar cell of its kind, nor is Caltech the only institution that has made them. Google "silicon nanotube solar cell" or "silicon nanorod solar cell" and you will get quite a few hits. Others have made photonic crystal-type solar cells from other materials -- some of those are potentially a very big deal because they can produce more than one electron-hole pair from a single photon.--Squirmymcphee (talk) 17:43, 11 March 2010 (UTC)Reply
I had another look at the paper over the weekend and noticed something I didn't pay attention to when I was skimming it for the efficiency figure: The rods the cells are made of are micrometer in diameter and at least 30 micrometers long. Thus, the structures might be a bit large to truly be a photonic crystal, at least at optical frequencies. Instead, it appears that the light-scattering particles they incorporate between the silicon rods produce plasmons, and these are responsible for the structure's enhanced light trapping. The Caltech group certainly has done a lot of work on plasmonic scattering in similar structures.
The other thing I noticed is that they say their device can capture 96% of incoming photons at a single wavelength and 85% of collectible photons over the course of a whole day. This is an important statement: Standard crystalline silicon technology can capture nearly 100% of incoming photons at a single wavelength and more than 85% of collectible photons, but only when the light is incident from near-normal angles. Over the course of a day, collection efficiency is significantly less. Of course, if the cell has to be put behind a sheet of glass to protect it from the elements then that advantage will disappear....--Squirmymcphee (talk) 14:16, 14 March 2010 (UTC)Reply

Tabbing and Stuff

edit

I miss asking you questions man. I hope spring is treating you well.

You've mentioned tabbing a few times. Can you suggest a primer on tabbing options? Do you see a likely course for tabbing to take in the future? What's the thing about tabbing that makes this step difficult? Are lasers going to simplify tabbing? Are thinner wafers going to simplify or complicate tabbing? I don't expect you to answer these questions entirely but I'd appreciate any pointers.

I've got a million PV theories going on in my head right now due to what I see as a lack of PV information inflow. Do you see a sort of unknown waiting thing going on in the PV community right now? Mrshaba (talk) 04:00, 24 March 2010 (UTC)Reply

Spring is good -- finally getting warm. Hope yours is going well too.
Not really aware of any good overviews on tabbing. I have seen it mentioned, almost in passing, in books such as the Handbook of Photovoltaic Science and Engineering, but my own knowledge has basically come through hands-on experience, discussions with manufacturers, etc. A recent Q-Cells paper comes to mind as something that might interest you. You might also Google the Crystalclear project, which is an EU project to create a PV roadmap that wrapped last year. Finally, I think Photovoltaics International has recently published an article or two that might touch on tabbing.
The difficult thing about tabbing is that you have long, narrow, thin contacts that have to be precisely aligned and soldered to the front of a cell, then to the back of its neighbor. While it is certainly not an insurmountable automation task, it is a complicated one involving a lot of handling (which is bad for yield). Thinner wafers will complicate tabbing not only because of the handling issue, but because they are more susceptible to bowing under residual stress left when molten metal cools (be it during contact formation in the solar cell manufacturing process or during soldering of tabs). I'm not so sure about lasers and tabbing -- to my knowledge, a bit of physical pressure on the tab against the cell during the soldering process is beneficial. On the other hand, with thinner cells maybe some sort of laser soldering would improve yield rates. It's not really my area of expertise, though I wouldn't bet on seeing laser soldering in PV anytime soon.
If you look at some of the things recently published in Europe (you'll probably find some of them on the EPIA web site) you will see an expectation there that a significant portion of the PV market will go to all back-contact cell designs in the next 3-5 years. This should greatly simplify tabbing, though the current sticking point is that there is no standard for the contact pattern, which prevents equipment manufacturers from committing to support such designs.--Squirmymcphee (talk) 11:49, 27 March 2010 (UTC)Reply
Thanks for the info. I'm pleasantly surprised to hear all back-contacts might make a go of it - I hadn't heard that. Here's a quirky article about solar power from the perspective of a printed circuit board guy. I'll warn you the article meanders and there are many errors but there are also some interesting non-standard analogies and info that you might enjoy. Mrshaba (talk) 07:57, 30 March 2010 (UTC)Reply
Interesting article. I'm surprised it doesn't mention ECN's approach to tabbing back-contacted cells, which is reminiscent of the way surface-mounted components are soldered to PCBs.--Squirmymcphee (talk) 12:49, 4 April 2010 (UTC)Reply

Hola Squirmy...

I was wondering how difficult it would be to design a solar cell that was tuned to a very specific range of wavelengths. Or you could think of it from the opposite end and shine a very specific wavelength of light on a solar cell. I'm wondering about short distance power transmission i.e in a room. Does this question make any sense? I'm wondering what the efficiencies would be? Mrshaba (talk) 14:55, 19 May 2010 (UTC)Reply

Hmm ... interesting question. Certainly with standard semiconductor technology you wouldn't be able to do it -- that responds to all wavelengths below a certain value. If you want to restrict the response to a certain range, you would have to filter out the short wavelengths that you don't want or use a bandpass filter. Outside of standard semiconductor technology, it might be possible with something like quantum dots, but that's getting a bit outside my area of expertise. Sorry....--Squirmymcphee (talk) 21:41, 24 May 2010 (UTC)Reply
OK... I was thinking there must be peaks in the responsiveness of some semi-conducting materials to certain wavelengths - sorta like absorption spectra. I was thinking you could someday build LEDs that emitted in this peaking range and couple them with cells that absorbed in this range. I'm thinking along the lines of: how do you get power to a hypothetical solar powered smoke detector on the ceiling, an emergency lighting LED on the ceiling, motion detectors, etc.
Do you think we are going to start seeing lots of hand held devices getting solar cells. Laptops, cell phones, remote controls etc.? I heard calculators mostly abandoned the solar approach not because the novelty wore off but more because batteries became cheaper - If this is true you would think the balance has shifted back to solar by now? Mrshaba (talk) 12:01, 25 May 2010 (UTC)Reply
There are certainly peaks in the responsiveness of semiconductor materials, but they are broad peaks. Incidentally, an ordinary LED can function as a solar cell of sorts. If you have a sensitive enough ammeter and light source, you can shine it on an LED and detect a current.
Hand-held devices: Yes, I think more of them will get solar cells. In a lot of cases I think it will turn out be a bad idea -- solar cells for the novelty, with common sense and good engineering lost in the shuffle. As far as calculators go, I can't say.--Squirmymcphee (talk) 22:21, 21 July 2010 (UTC)Reply
One other question for you. MEMC recently bought Solaicx and I read this quote about the acquisition, "This type of technology seemed brilliant when silicon was at record-high market prices a few years ago. It's a little less attractive when prices are low..." I've heard the same exact thing said about FBR plants but this puzzles me. I would think technologies such as FBR are actually in a better position (relative to their competition - Siemens) when silicon is cheaper. Any thoughts? Mrshaba (talk) 13:25, 25 May 2010 (UTC)Reply
A 1% increase in efficiency saves X number of dollars in installed PV system cost (where the value of X depends on who you ask). By using lower-grade silicon you save money, but take a hit in efficiency. When quality silicon is expensive, the money you can save by using UMG and such is greater than X, so it makes economic sense to use low-quality silicon and take the efficiency hit. When quality silicon is cheap, the money you save is less than X and it makes more sense to pay extra for the good stuff. FBR is still a bit of an unknown quantity at commercial scale, so the attitude among many is "why take the risk when we know we can get the good stuff cheap?"--Squirmymcphee (talk) 22:21, 21 July 2010 (UTC)Reply
Yet another one for you. What do you think is the practical limit for the lifetime of panels? Mrshaba (talk) 17:56, 19 June 2010 (UTC)Reply
Tough question. In principle, the cells themselves should last many, many decades. The issue is keeping them, and the other module components, safe from the elements. Right now the limiting factor tends to be encapsulant, which can produce acids that corrode interconnects and degrade cells over time. I would like to think that the practical limit is exactly the same as for the glass itself, but I don't know if that's realistic.--Squirmymcphee (talk) 22:21, 21 July 2010 (UTC)Reply

I talked to an Apple engineer about putting solar on laptops, iPhones, etc. He said it was a heat concern. I can't say I completely understand why but then I thought about the unintended consequence people putting their solar laptops or iPhones out in the Sun to charge. That would be bad. Then again, right now my cell phone is sitting on the table. If it was getting a trickle charge from the ambient light that would be nice. Having a self-contained recharge function would also be nice on those occasions when you forget to bring your charger on a trip. The Apple engineer said they wouldn't put solar on laptops until it was a dollar a watt. I told him First Solar is under a dollar per watt (cost) and there's probably some a-Si that is about $1/Watt as well. He was surprised. I'm surprised he had a number in mind at all. Mrshaba (talk) 19:52, 21 July 2010 (UTC)Reply

Thin film would be a good call there, because of its good low-light performance. And Apple would like it because they have cells made in all sorts of funky shapes and colors (at least in principle). As you say, it would be good for a trickle charge, but even in bright sunlight you won't get more than a few watts out of a thin film panel the size of a laptop....--Squirmymcphee (talk) 22:21, 21 July 2010 (UTC)Reply
Thanks for all your thoughts. Can you offer any kind of guess as to how average panel lifetimes might change over the next 10 years? Is this something that is looked at much? I could see how it wouldn't be but I'm curious. Mrshaba (talk) 01:29, 22 July 2010 (UTC)Reply
My sense is that there is more thought being put into recycling of panels than there is in extending their lifetimes. A lot people in the industry feel that a majority of today's panels will already last 35-40 years or more. When you're talking about those kinds of lifetimes, you're reaching a point of diminishing returns in the levelized cost of electricity. For rooftop installations, you're also talking about people having to remove their panels to replace their roofs at some point, and if the panels are already quite old then people might just replace them anyway rather than risk having to go through whole uninstall-reinstall process with new panels a few years later. That's not to say that nobody is thinking about extending panel lifetimes -- if you can reasonably expect a panel to last 50 years then you'll keep through a re-roofing job, for example -- just that I don't think it's a very high priority for many folks. Of course, if you want to talk novel packaging, certain thin-film designs, or organic or dye-sensitized panels then it is a different story. Those areas are quite active when it comes to extending panel life.--Squirmymcphee (talk) 11:48, 7 August 2010 (UTC)Reply
"When you're talking about those kinds of lifetimes, you're reaching a point of diminishing returns in the levelized cost of electricity." This is exactly why I asked the question. I've been modeling 30 year lifetimes but I wanted to get a feel for what kind of product is currently out there - I feel more comfortable modeling 35 and 40 year lifetimes now. As an end-user you'd have to compare the extra costs associated with extra lifetime and pick your optimized trade-off point. Each customer is going to be different though so manufacturers can't design their product to a specific trade-off point. I suppose they just need to build a good product that has a lifetime comparable to the rest of the industry. Thanks for the info. Mrshaba (talk) 23:07, 8 August 2010 (UTC)Reply
As far as FBR poly goes my thinking is that cheap is not good enough. Not when a few cents per watt start to matter. But I agree with your thoughts about the perceived riskiness of FBR. I just think we'll find out the perceptions are wrong. MEMC is becoming a force to be reckoned with. I figure they'll announce another poly project or large expansion within the next few months. We'll see. Mrshaba (talk) 01:36, 22 July 2010 (UTC)Reply
One thing I neglected to mention is that the drop in feedstock prices last year slowed the drive to thinner cells -- in fact, some manufacturers went back to slightly thicker cells to improve their yields on older production lines. This really hurt UMG producers because you lose less efficiency in the switch to UMG when the cell is thin versus when it is thick.
I suspect FBR will turn out just fine in the long run, but everybody wants somebody else to be the guinea pig before committing to it. MEMC is in a good position to promote it, since they also make their own wafers. The risk is taken off the cell manufacturer who buys their wafers, as opposed to a situation where a wafer manufacturer (or a cell manufacturer that makes its own wafers) has to choose whether to buy Siemens- or FBR-produced feedstock. It will take time, though -- feedstock plants take a lot of time and money to build, and the vast majority of those under construction are still Siemens process.--Squirmymcphee (talk) 11:48, 7 August 2010 (UTC)Reply
I agree with your thoughts about FBR for the most part. Again, I figure when Si prices are high the developer wants to get a facility up and running as fast as possible - they don't care if the Si costs 5$/kG more to produce, they care about the project timeline. When Si prices are comparatively lower developers care about those $5/kG a lot more so FBR becomes more attractive. I realize there will be lower overall interest in Si when prices are lower. Mrshaba (talk) 23:07, 8 August 2010 (UTC)Reply
Remember when I suggested incentives would tend to float PV prices? I built a basic LEC spreadsheet to test this theory. I came up with very consistent ROI figures for 2006 through the beginning of 2009. Since then, the ROIs have been considerably higher - not surprising when you consider the drop in panel/system prices. The float theory doesn't predict this discontinuity - it would only predict that prices will respond to the existing ROI conditions in such a way as to return the ROI to a stable floating range. I don't think prices in Germany have reached a stable range yet. Accordingly, I figure prices in Germany will stay flat/trend up slightly for the rest of the year despite the FiT adjustment. Given the current degression schedule I'd also expect a minimal drop in prices going into next year. This suggests the Germans did a pretty good job adjusting their tariff. Mrshaba (talk) 23:07, 8 August 2010 (UTC)Reply

PV simulator

edit

Hi Squirmy...

I'm looking for an excel spreadsheet type that models hourly PV performance based on input meteorological data and PV system design variables. I'd like to see the data in hourly increments. I know such excel based PV models exist but I can't find one with all the coding unwrapped. Do you know of one? Do you know who I should ask? Seems like all roads lead back to Sandia. Mrshaba (talk) 19:29, 13 August 2010 (UTC)Reply

Honestly, I think Excel is the wrong tool for the job. That said, the Canadian government did a pretty decent job with its RETScreen tool, though last I checked the coding was not "unwrapped" as you would like. I do not recall whether it shows simulation results hour-by-hour and I do not currently have a copy on my computer. The tool I rely upon most often nowadays is the Solar Advisor Model (SAM) from NREL. The underlying PV system simulation is based on a propriety code from the University of Wisconsin called TRNSYS, so you definitely will not be able to see the code "unwrapped". However, TRNSYS is well validated and the PV-specific models that SAM applies are publicly documented. SAM itself is not open source, but it is freely available from NREL and the hour-by-hour results can be exported to Excel. (It also includes a utility that will plot the hour-by-hour results for you -- poorly, in my opinion.) Be warned: SAM has a lot of inputs -- if nothing else, it will give you an appreciation for just how important it is to evaluate PV on a case-by-case basis (e.g., for two neighbors in Southern California, PV may be economical for one and not the other solely because one can get a mortgage and the other cannot).
I am aware of a project to create a sort of open-source version of SAM -- it would use many of the same models TRNSYS and SAM do -- but I do not know whether it will ever get off the ground. If it does, I will mention it to you. There, all of the code will be available, but it will be a cross-platform stand-alone application (and therefore definitely not be coded in Excel).--Squirmymcphee (talk) 17:45, 14 August 2010 (UTC)Reply
Thank you for all the info. For now I'm using Excel because I understand how it works and I'm pretty sure I can write the PVFORM based codes required to do what I want it to do. Right now I'm just collecting the equations and trying to find the linear cheats. I've looked at SAM and compared some of the results to what my model comes up with. The Levelized electricity costs are identical but for some reason the NPV values are well off. Apparently there are some glitches in their financial modeling so I'm holding my breath for the corrections - next release in September. Thank you so much for the info. Cheers. Mrshaba (talk) 18:48, 14 August 2010 (UTC)Reply
I managed to crank out an Excel based version of PVFORM. It's a 20 MB beast at the moment and it takes 10 to 15 seconds to process a location but it works. I've only tested out 6 stations so far but I'm within about 5% of PVWATTs on 5 of them. Couldn't really ask for more. I think the hardest part of the whole process was dealing with PVFORM's wonky astronomical algorithms - head banging fun that was.
I found the issue between my economic performance model and SAM's. I was using a user chosen nominal discount rate by default whereas SAM calculates this value based on a real discount rate and inflation. SAM's method is unnecessarily complicated if you ask me. Anyways, once I corrected for this discount rate issue I found my results spot on with SAM's. For my next trick I'm going to incorporate some self-consumption modeling and tie everything together with an NPV optimization algorithm. Mrshaba (talk) 13:51, 28 August 2010 (UTC)Reply

I made an additional translation of the PVFORM code from Fortran into VBA. It still uses an Excel based interface but it's running above Excel now. I haven't worked all the bugs out yet but I'm very close. The code is compact and very fast. I've been in touch with the licensing people at Sandia about getting permission to share/distribute the code. I've also been in touch with Dr. Menicucci who thinks I should bypass the licensing people altogether because the equations I'm using are all in the public domain. Dr. Menicucci is an absolute riot. I was wondering if you could tell me who might be working on this open source version of SAM? Is it physics based like TRNSYS or empirical like PVFORM? I'd like to take a look at what pieces they have so far or find out the direction they are thinking about going in. Maybe even bounce the PVFORM based model I've been working on off them and see if there are any suggestions for improvement? Cheers Mrshaba (talk) 19:06, 20 September 2010 (UTC)Reply

SAM's method of handling the discount rate does look unnecessarily complicated, I agree, but I think you will find that it makes a significant difference when the nominal discount rate is high and (especially) when the inflation rate is high. (I'm making that statement from memory, so don't hold it against me if I got it slightly wrong.) If you have not already checked out this manual I highly recommend it.
I hate VBA with a passion, so I have to hand it to you for translating PVFORM into that wretched language. I agree with Menicucci -- there is nothing proprietary about the models or equations, and as far as I know PVFORM does nothing magical in its algorithms. There was a Ph.D. thesis at Georgia Tech a few years ago by a guy named Alex Pregelj who, as I recall, re-wrote PVFORM in Matlab, though he replaced one or two of the equations with improved versions. Don't know if the code was ever released to the public -- maybe the source is in his thesis? You might be able to download that from the Georgia Tech library.
I know very little about the would-be SAM clone, and as far as I know it is just an idea -- I only know about through a common contact. As I said, it may never even get off the ground....--Squirmymcphee (talk) 16:57, 22 September 2010 (UTC)Reply
I'll look into refining my use of the discount rate. Truth is, I was just happy to find the discrepancy and haven't touched the economic model since. I used that NREL manual you linked to as a primary source for my LCOE model. Like I said, I get the exact same financial results as SAM. Funny how that works when you use the same equations.
Everyone agrees with Menicucci concerning PVFORM except for the bureaucracy at iPal. They're reviewing my versions of PVFORM now. If need be, I'll get their Free-use license just to settle the situation.
I've been looking around the Georgia Tech site recently but I haven't come across the Pregelj paper. Another Georgia Tech alum, Ristow, mentioned they had a local PVFORM based model called GeTPV with an improved Thermal component. I've been trying to track down the model without much luck. The Pregelj program probably references the GeTPV model if it isn't the GeTPV model itself. Thanks so much for the lead. As far as an open source SAM project goes, I'm going to go ahead and put a TRNSYS based flat plate model together myself. It will be an adventure. The end product should be relatively close to SAM. Mrshaba (talk) 22:07, 22 September 2010 (UTC)Reply
Hola... I've been studying semiconductor fundamentals lately in the hopes this will help me with my PV modeling. I've been reading Donald Neamen's Semiconductor Physics and Devices and various papers (Wolf, Loferski, etc). It seems that Singh and Sze keep coming up. Would you recommend any of Sze's or Singh's textbooks to someone looking to bone up on their fundamentals? Are there any other textbooks that you'd recommend instead? Something more specific to PV perhaps? Mrshaba (talk) 01:12, 12 October 2010 (UTC)Reply
I'm not familiar with the Singh book -- from what I gather it was published after my initial introduction to semiconductor physics, and since those fundamentals don't evolve much I have had little reason to upgrade my library. I learned from a series of books by Pierret et al., which I believe have now been combined into a single volume. Sze is a great reference, but I don't recommend it as a book to learn from -- he assumes the reader already knows quite a bit. The Pierret book(s) have been undergraduate-level standards at many universities for quite some time, though I gather that more recently the Singh book has been very successful. As far as PV-specific semiconductor fundamentals go, I find Jenny Nelson's The Physics of Solar Cells to be very readable and an excellent introduction for solar cell purposes. If you want to go deep into the subject she will not get you there, but as someone who works with PV-related semiconductor fundamentals every day I still find the book useful once in awhile to refresh my memory on various topics. Martin Green's Solar Cells is also readable and provides a good introduction, but it was written so long ago that it does not cover certain topics that Nelson does (e.g., thin films).--Squirmymcphee (talk) 13:21, 24 October 2010 (UTC)Reply
Would you recommend any of the books in this reading list. Apropos of nothing, you might be interested in the Tipping Point. It's a quick and entertaining book about the contagiousness of ideas and whatnot. As I read it I wondered about the tipping point for PV. You might like it. Mrshaba (talk) 01:32, 14 October 2010 (UTC)Reply
I really dislike Stuart Wenham's Applied Photovoltaics -- I think it is poorly organized and poorly written -- but then I have the first edition. I have skimmed the latest edition and it looks like a big improvement, but I cannot say more than that. I am not familiar with the device physics sections of the Tomas Markvart book, so I cannot comment on them. The Luque and Hegedus book is a compilation of things, with each chapter written by a different author. It is a good in-depth introduction to a lot of topics, including device physics. I have a good impression of the first edition of the Fonash book, but that is all I can remember about that one -- I don't actually own it -- and I have not seen the latest edition. Finally, I am not familiar with the Goetzberger book on this list, but a previous book of his -- Crystalline Silicon Solar Cells, I think is the title -- was excellent in my opinion.
I am familar with The Tipping Point. I began reading it a few years ago, but I got sidetracked by something else after about 25 pages and haven't returned to it yet. It's still sitting on my bookshelf, though, so I'll get to it one of these days....--Squirmymcphee (talk) 13:21, 24 October 2010 (UTC)Reply
Any comments on this GTM article? Mrshaba (talk) 17:30, 20 October 2010 (UTC)Reply
Yeah, I'm not that impressed. Ion implantation is slow and expensive, and despite what Varian says it is not a low-temperature process -- damage caused by bombarding the silicon with ions needs to be annealed out. Varian is trying to make it faster for the PV industry, but even then the cost and the limitations of the technology itself will ensure that their only potential customers are those using single-crystal silicon to make high-efficiency solar cells, at least for the foreseeable future. The limitation in the technology itself is that implantation works very poorly on multicrystalline substrates -- the depth of penetration of the incoming ions is highly dependent upon grain orientation, so to get a nice, uniform emitter you need single-crystal silicon. The real advantages of ion implantation, at least as the technology stands now, is that it makes patterning easy and it makes it easier to work with certain dopants, like boron. It might be attractive to a company like Sunpower, which has heavily patterned n- and p-type regions all on one side of the cell, and maybe in the future it will prove effective on a wider scale for making selective emitters (which is what Varian is pushing). However, for most of the industry my suspicion is that it will prove too slow and expensive without a lot of additional work on Varian's part.--Squirmymcphee (talk) 13:21, 24 October 2010 (UTC)Reply

Thank you for the thorough replies to my random questions. You are my sole first person connection to the PV manufacturing landscape. You have always been a wonderful sport. One of these days I hope I will be able to answer some of your questions or perhaps lead you in the right direction as you have done so many times for me. Anyways... I finished off a rough draft of my SAMish PV performance model - I've named it PEP for Photoelectric Performance Model. All the formulas are in and the databases are built. The contraption is wonky at this point - half built in excel (the financial model) with the PV performance model built in VBA. The fire of creation has died for now but I don't quite have it up to snuff. By snuff I mean usability. Could you share that contact that wanted to build an open source SAM by chance? Via private wiki email maybe? Meh... One day of "wake up and do it" and I can get it going but I'd really like to bounce what I have against some smart folks before release. I have a rough engine at this point but I'd like to tune it up and make it more usable before releasing it. I've come this far alone and I can go farther but additional eyes will only help. You've mentioned that you know folks (indirectly) that are interested in this problem. Mrshaba (talk) 05:24, 4 November 2010 (UTC)Reply

No pressure... I can probably bug the SAM user group. Mrshaba (talk) 19:52, 4 November 2010 (UTC)Reply

Goofy Ideas

edit

Any idea what the fascination is with things like CPV and companies like Solyndra? Do these technologies make you giggle or do you think they deserve to be taken seriously? Seems to me that even a reasonably good contender like CIGS is given too much attention considering the rate at which crystalline silicon is expanding. Meh... I guess it's just the press I read. Have you heard anything about this projected glut in the first half of 2011? Mrshaba (talk) 23:19, 26 November 2010 (UTC)Reply

Other Ideas

edit

Here's one I don't think I've ever asked you about. The other day I met with a VC fellow involved in solar. We batted stories back and forth. It was fun. He mentioned that one of the problems that larger solar production facilities are having is that there are dozens (I can't recall the actual number) of different wafer bins - i.e. one bin for X efficiency, another for X + .2 efficiency etc. I don't know what all the causes are for this effect. I thought one of the causes might be larger ingots. Shear size and throughput probably come into play as well. Can you offer any insight into this?

I've been reading here and there about cell level MPPT. I find this interesting but don't know much about the costs. If you did have a cell level MPPT feature wouldn't you be able to mitigate some of the quality control issues mentioned above? The VC fellow thought (and I agree) the better solution is to go to the root of the problem and eliminate the variance in efficiency as much as possible via quality control and process measures. He named companies that are concentrating on this and pointed out that you can stack cell level MPPT on top whatever type of cells you have and it should improve performance. Do you think cell level MPPT is likely to be competitive? How far off do you think this technology might be?

All is well in my neck of the woods. I'm having fun running my power plants. Knock on wood. Hope all is well with you. Happy new years. Mrshaba (talk) 14:35, 19 January 2011 (UTC)Reply

Hey there, long time since I checked into Wikipedia and I'm on my way to dinner so I'll be quick. Yes, we (and every other cell manufacturer I know) have lots of efficiency bins. They're based primarily on output current rather than efficiency, though, because the purpose is to minimize losses due to mismatch between cells (which you do by ensuring that all series-connected cells have the same maximum power point current, since they all must have the same current running through them). I have never seen a cell-level MPPT proposal so I cannot comment other than to say that if it is not something integrated into the cell then I do not see it happening -- it seems to me that you would have an enormous reliability problem there, even if the cell-level MPPTs were individually 99.9% reliable. In contrast, I think variances in outputs can be reduced substantially -- that is something the PV industry is only beginning to look at, and believe me, there is a lot of room for improvement.--Squirmymcphee (talk) 16:44, 19 February 2011 (UTC)Reply

Silver question

edit

I did some simple BOAT calculations today and determined that silver is getting to be a rather large portion of production costs. I thought it was you who had mentioned that every PV company had a silver reduction plan but scanning through the page here I can't find that quote. I figur PV companies have hedged silver contracts so they aren't hurting from silver spiking just yet but they're probably more concerned about silver than they were a year ago. What are the main alternatives to silver? Lead? Copper? What are the tradeoffs that you suffer when you switch out silver? In your opinion, how will silver consumption get trimmed and/or replaced?

Hope your spring is springtaculous? Mine... Not so much but at least the hockey club is doing well.

Best Mrshaba (talk) 21:47, 29 April 2011 (UTC)Reply

It's probably more accurate to say that silver paste manufacturers have hedged silver contracts. There might be some PV companies that have long-term contracts of some sort with paste manufacturers, but paste technology moves rapidly enough that it is not unusual for companies to change pastes a few times a year. Some even have a "basket" of pastes that they have validated for production and they simply switch between them on based entirely on which one happens to be the cheapest when it is time to re-order.
At any rate, you're correct that silver costs are big. So are aluminum costs, at least for standard technology (where the entire back side of the cell is covered in aluminum), though that is primarily because of the quantity of aluminum used rather than the unit price. Lead is totally out as an alternative to silver. It is not conductive enough, though it is used in current silver pastes in its oxide form as a sort of catalyst for good contact formation, but it does not actually carry any current itself. Even in that form, it is being phased out thanks primarily to European RoHS laws. Copper is the likely long-term replacement, but in the short term silver will probably just be used more judiciously. Reductions of 50% or more in silver consumption should be possible through electroplating the contacts instead of screen-printing them. Copper contacts would also be plated, but they will first require the development of a low-cost, high-throughput barrier layer between the copper and the silicon to avoid efficiency-killing contamination of the cell by the copper. Plated silver contacts should perform better than screen-printed ones, and copper contacts better still. The tradeoff, then, is the more complex and expensive manufacturing process.
Similarly, aluminum costs can be cut by as much as 90% by evaporating or sputtering thin layers instead of screen-printing thick ones, but this requires solar cell designs that do not rely on having a thick aluminum layer on the rear. These designs have long existed in the laboratory and currently exist, to a much lesser degree, in production as well, though they are far from dominant in production yet. The tradeoff here is that evaporation or sputtering require huge, expensive vacuum chambers, which require significantly more upfront capital (in exchange for dramatically reduced operating costs).
See if you can find a copy of the International PV Roadmap by SEMI. There you will find some information on how European PV manufacturers believe metal consumption will evolve over the next decade.
Spring has been good, unexpectedly warm. I have the vague impression that by "hockey club" you mean the San Jose Sharks; if so, one of us is not going to be happy in a couple of weeks....

--Squirmymcphee (talk) 23:05, 29 April 2011 (UTC)Reply

Thanks for putting all these ideas together. I realize that silver goes up and down but the more troubling thing at work here is that PV is consuming a non-trivial amount of the world's silver supply so further growth will put more pressure on Ag prices than previous growth has. This isn't new to you I'm sure but I hadn't realized the magnitude of the situation until yesterday when I sketched things out. What I'm wondering here is how close the equipment makers are to supplying equipment at an attractive enough price to entice manufacturers to jump over. You'd figure the manufacturers must be relatively close because this problem has been on people's minds for some time. Then again, the equipment makers have to know they've got leverage here. I'm thinking there's going to be a multi-year long silver kerfuffle.
Just to get something straight... It seems as though the drive to reduce Al fights against the desire to reduce Ag. Will one get replaced/reduced before the other or will things run parallel.
A warm spring you say? Now you're just being mean. Canucks are going all the way this year. Mrshaba (talk) 22:08, 30 April 2011 (UTC)Reply
Funny how silver prices have changed so quickly in the last few days. The certain fate of the Sharks as well... Coincidence? Hmmm... You might enjoy this article. Finley argues that the oversupply situation might spur a transition to the next generation hi-eff manufacturing equipment that I for one have been anxiously waiting for. We'll see. Mrshaba (talk) 06:19, 11 May 2011 (UTC)Reply

I read through the SEMI report you suggested. Lots of interesting info in there. One thing we talked about above that was addressed in the report was the expected move to fully squared mono-wafers around 2015. They also mention the move to back-contact designs and silver-less designs over the next 5 to 10 years. Speaking of silver they set the current usage rate at .3 grams/Watt which is 2.5 times higher than I expected. Anyways, you've already seen it. Just saying thanks for recommending it to me. The Canucks indeed went all the way but they couldn't get that last Win. Oh well... Mrshaba (talk) 18:27, 28 June 2011 (UTC)Reply

Wafers

edit

What do you think of this new ingot technique? Mrshaba (talk) 13:59, 18 July 2011 (UTC)Reply

Not sure what to make of it yet. Different manufacturers are encountering different issues with it, and it wouldn't surprise me if it works well for some and not for others. I know of one company, for example, that up to now has only made multicrystalline wafers, so they have buy seed crystals from mono manufacturers if they want to make the mono-like wafers in the article. Another manufacturer loses the crystal structure part way through the ingot, so each ingot produces about 1/3 near-mono, 1/3 mono/multi blend, and 1/3 multi. And so on. My company is one of many that will be experimenting with it, and I will be very interested to get my hands on some.--Squirmymcphee (talk) 23:03, 24 September 2011 (UTC)Reply

It's Fall

edit

Hello Hello... Apparently it's fall... Football and Hockey are back at it. That makes me happy.

I am rather confused by the state of the industry at the moment. Do you have any thoughts?

Here's a wafer question for you... I don't know how familiar you are with 1366's direct wafer process but can you offer a guess as to how large these direct wafers can be made? Crazy idea #286, why not make the wafers the size of a whole module and laser scribe the cells, vias etc. Have you heard this idea put forward before?

Hope all is well. Mrshaba (talk) 02:30, 22 September 2011 (UTC)Reply

I'm glad football and hockey are back too, though I've been in Europe and haven't gotten to see any yet....
Being inside the PV industry makes it no less confusing, believe me. Every day I show up at work half-expecting to find out we've been bought out by somebody (and there's not even a particularly good reason to think that might happen). SEMI reported this week that equipment orders are down, which I think is a good thing in the short term because of overcapacity in the industry, but it looks like the second half of 2011 will be better than the first in terms of cell and module shipments. That's not so unusual, I think, but the first half was pretty crappy this year.
I am somewhat familiar with 1366's direct wafer process, but I am party to an NDA with them. Because of that, even though I have not done anything directly related to their direct wafer process, I won't discuss anything that you cannot find on their website or elsewhere online -- sorry.
As for making wafers the size of a whole module, I can think of a lot of technological hurdles that would have to be overcome to make that work. First off, the cross-sectional area of an ingot is typically significantly less than that of a module. Our suppliers typically cast ingots that yield 16 bricks from which wafers are cut, meaning that you get a 4x4 cell area from a single ingot -- only enough to produce 8-9 volts. Ingots are growing, so this could change in the future, but at the moment you would have to use multiple wafer slices to make one module. (Note that through all of this I am assuming cast multicrystaline silicon -- if you're talking mono, you have much, much longer way to go before ingots are that large.)
Next, there's the wafer slicing. The larger a wafer, the more likely it is to break during slicing. To slice such a large wafer, you would at the very least have to make the wafer thicker and/or reduce the slicing speed (and therefore throughput). Most likely, module-sized wafers would require entirely new wafering technology.
Assuming you make it that far, you have a couple of major processing issues. First, you need equipment that can handle such large wafers. That in itself is not trivial. Then you need to ensure that equipment is gentle enough not to break the wafers, which rapidly becomes more difficult as wafer size increases. A broken or cracked wafer in this setting would be a huge loss -- an entire module's worth of cells, essentially, because automated equipment generally does a lousy job of handling damaged wafers without causing further breakage. In other words, you would be creating a setting where maintaining or increasing yields from existing levels is not only more crucial than it is now, but also far more difficult.
Finally, when all of the cells are finished an isolated, you may find that they are not suitable to be connected to one another. In general, the cells from the center of the ingot will have higher currents than those from the edges, which will in turn have higher currents than those from the corners. Therefore, by interconnecting them, you will be giving up quite a bit of power. Most likely, they will have to be sorted into bins. And if you're going to do that, there isn't really much point in processing an entire module's worth of cells as a single wafer.
All that said, I can see somebody trying your suggestion someday. That day will probably not come soon, but if the wafering technology and the processing equipment to accommodate such large wafers ever exists, I'm sure somebody will give it a go. Perhaps by then, quality control will even have progressed to a point that sorting is no longer an issue....--Squirmymcphee (talk) 23:46, 24 September 2011 (UTC)Reply
Definitely a wonky market. When we look back on how much got installed this year people are going to be quite confused.
I realize the module sized wafer idea is out there. I was thinking more along the lines of getting these large sheet sized wafers with a non-traditional process like that used by 1366. You bring up a lot of hurdles you'd have to get over. The breakage problems had occurred to me but the issue of the crystal quality gradient hadn't. This brings forward a related question... Won't these monticrystalline wafers lead to a lot more bins? Mrshaba (talk) 05:00, 25 September 2011 (UTC)Reply
Heh, "monticrystalline" -- by that I assume you're referring to the mono-like wafers, but I like that word! Anyway, I have no idea at this point how they will affect binning, but I suspect that they're likely to more heavily weight the higher-efficiency bins in existing distributions. Expanding the number of bins would not be desirable, but it is also not the end of the world and is rather expected when you know you're receiving material with a wide quality variation. Long-term, of course, quality control efforts would be made to rein in the variability, but it's simply not going to be as easy to do as with plain mono or multi. Nothing comes for free, and that's the tradeoff for filling the higher-efficiency bins at lower-efficiency prices.--Squirmymcphee (talk) 16:50, 29 September 2011 (UTC)Reply
Yes... Mono-like wafers... Glad you like my wordvention. As always, I appreciate your thoughts on binning.
Last night I reread an editorial by Khosla that talked about the difficulties for thin-film start-ups. He laid out what he saw as targets that TFs need to hit. I generally don't think about thin-films because standard silicon provides a more predictable picture of the future. But then I read Khosla and it got me musing. I've probably asked you this already but I'm asking again: Do you see any serious competition coming from a thin-film technology any time soon?
One theory I have is that somebody (First Solar is my guess) is going to snatch up a lot of patents and collect a bunch of talent during the next few years. This new collection of talent could lead to a crystal killer. I don't think the odds of this are all that long. Any thoughts? Mrshaba (talk) 17:11, 30 September 2011 (UTC)Reply

Non-Wafers and Industry Growth

edit

Quirky question #283. In Amorphous silicon processing are the surfaces that Si forms on textured? I know that water vapor prefers to crystallize on surfaces that mimic the crystal structure of ice. My general understanding is that the surface forms the seed pattern for the ice to plate out on. I've been wondering if there might be a way to convince silicon to plate out in a multi-crystalline or mono-crystalline versus amorphous pattern by surface texturing? The first crude idea that came to mind was to laser scribe the glass with a silicon nucleating pattern and plate silicon there. Another was to scribe the back surface. Another idea was to use a coating to form your mimic pattern. What kind of coating I don't know. My searches for silicon nucleators come up with no leads.

These are all very fresh brainstormy ideas. Just throwing them out there because I can't recall reading about anything quite like this. Do you know of anything similar? I'm just thinking that it would be wonderful if a process came along that formed a crystalline surface without having to go through all those pesky processing steps. Duh right? I can't help but think that something is going to come along that makes all the wafer, ingot and cell processing lines obsolete in a very short time. The rags are lately making a big deal about China taking over PV but a large part of me thinks the Chinese players are going to be left holding an empty bag that they've spent tens of billions on. Crystal is the best bet today but the Dragons would all be screwed if a wafer-less process came along in the near term. Do you industry guys talk at all about how something new could come along in PV that upsets things? I hear more and more people talking about PV becoming a trillion dollar industry - this implies 4ish more doublings of market size and around 5 doublings in annual shipments. You'd think this sort of size implies that a lot more brain power is going to come into the field. At some point PV will become the Big Game that all the cool kids want to play. I'm not saying the field isn't already full of geniuses - just saying that a I could see a new flux of talent pouring into the field over the next decade or so. It makes sense to me that this will speed up the rate of progress and perhaps lead to some Eureka moments. Any thoughts? Mrshaba (talk) 18:02, 29 October 2011 (UTC)Reply

Are these guys overstating at all? It would seem that if OCF can help with controlling impurities then you might be able to use lower grade poly? Mrshaba (talk) 23:08, 29 October 2011 (UTC)Reply

Wondering

edit

Anything new and interesting? Mrshaba (talk) 03:23, 14 July 2012 (UTC)Reply

Average Module Efficiency

edit

Hi Squirmy... I was wondering as I always am... Do you know of a resource that lists the historical trajectory of average cell/module efficiencies? I've got all the Champion cell/module data from Green's reports and the NREL graph but I only have average data for modules going back to 2003. It would be interesting to have a sketch of data points going back to the 50s. Mrshaba (talk) 03:59, 21 August 2013 (UTC)Reply

Multiple Junction Cells

edit

Hi Squirmymcphee... I've been reading all the headlines about perovskites lately. The commercialization of this technology has gone from meh to hmmm to seemingly inevitable in a short amount of time. I've read that perovskites could be stacked on top of silicon. Do you think this is realistic? Would it be possible to stack perovskites on top of perovskites that are tuned to different parts of the spectrum?

Mrshaba (talk) 19:21, 29 July 2014 (UTC)Reply