Anyone here have any educated guesses how much energy it takes to build a square centimeter of photovoltaic cell? The real question comes down to how long it takes for a cell in a place on Earth with high insolation (high average sunlight) to pay back the energy it took to make that cell. Associated with that, what it the expected service life of a photovoltaic cell in normal sunlight at the Earth's surface? I suppose all this depends on the cell technology, but the answers could help pick an optimal technology.
This "typical" energy payback time is misleading (in that it does not appear to take into account the energy required to pull the resources out of the ground to make the system, payback time for additional required system equipment such as the batteries, regulator, goverment discounted pricing incentives, etc).
Payback time is much, much too low.
Solar panels have there place, but at the moment the only place I see them being useful is in powering remote applications where I have no access to a mains supply.
Where is our resident solar guru when we need him !
PV cells are 6" silicon wafers like integrated circuits are made from... the pure crystal of silicon is 'grown' from a vat of molten silicon in an electric furnace about 1500 deg C. Then the cylindrical ingots are sawed into wafers. The cells are big diodes, so a 'boat' of N type wafers will get put in a diffusion furnace at 1500 deg C for some precise number of minutes with a P type gas blowing across it. Then the PV cells are assembled into panels. You can see that both processes are energy intensive. There are only a few companies that make the wafers.... IC manufacturers have the wafer fabs to make ICs, so there is a supply and demand force at work on the silicon itself. If a 150 watt PV panel sells for $750 (about $5 a watt), you can bet that the cost of the electricity to make it plus some profit is in that price.
Whatever happened to "Edge-defined film growth"? That was supposed to solve the problem of sawing the ingot, because it grows a very thin ribbon of single-crystal silicon. But I haven't heard of it for quite some time. Anybody even ever heard of such a thing?
Hmmm...not necessarily. It may be that the semiconductor manufacturers have already paid for the cost in buying the high grade part of it, and the "castoffs" used to make the PV cells are being sold below the cost of the refining energy, but above the value as scrap. If YOU had a big pile of something left over that didn't meet the requirements of your prime customer, but had little intrinsic value, and you found a customer willing to pay something above the scrap value for your scrap, what would you do?
Of course, not all the PV cells are made that way. I suppose it will only be when (if?) people get serious enough about using PV as an electrical energy source that we'll be able to tell for sure just what the real cost is, but it seems like most sources right now are saying that's it's around four to five years to pay back the energy, with some reasonable belief that it could become shorter. Even at four to five years, it appears that it's not out of line with the cost per watt of other facilities built to produce electric power.
PV isnt a commodity yet evidently. The price seems to be about $5 a watt, because everyone is selling as much as they can make at that price. Others on this newsgroup have said that about the time some warehouse manager starts complaining about 'all those solar panels from last year filling up the warehouse... move em out!' Then the price will start coming down. There is some Dept of Energy paper that says they take about $3 a watt to make, so 40% profit right now. That theoretical
150 watt panel that cost $750... at $.15 per KWH it would need to put out 1000 KWH to pay for itself. .15KW x 5H a day is .75KWH a day for
1333 days... about 3.5 years. Another back of the envelope sketch by a master!
The interest/expected ROR on that $750 panel is about $.20/day (10% of $750), so the first 8ish hours of sunlight are the bank's. How many hours of sunlight do you get on an average day?
========================================== I've heard they put out a little less when hot, so maybe a concentrator would work well in Maine, but not so good in Arizona?
They do not work well in the shade. And in bright and sunny conditions, a mechanical generator might be more efficient and a whole lot less expensive per watt.
A different perspective is teh allowed size of the panel.
Monocrystalline cells may have high efficiency but are expensive.
Thin film on glass may be less efficient, but if space is no issue the size doesn't matter and cost to output or energy expenditure to output could matter.
What is desirable on a satellite is not the ame as on a residential rooftop.
There really isn't any scrap involved. There is a finite amount of silicon being made, all existing silicon manufacturing plants are running at full capacity. Most of production goes to semiconductor manufacturing (which pays a higher price for higher purity and more precise doping - exact same raw material with further processing by the material mfgr.) and essentially all of the rest goes to solar cell production, which is currently limited by inadequate silicon supply. The silicon manufacturers are however reluctant to add capacity, remembering as they do the last industry downturn and figuring that the next one will coincide with any new capacity coming on line. So you should not expect to see any drop in solar cell prices until the next major semiconductor industry downturn.
This is per one of my sisters who has been selling solar power systems for over 20 years. She also claims that grid connected solar only pays back the investment per standard accountant analysis if it is used for long term emergency backup power instead of a diesel generator set, otherwise buyers are either making an environmental contribution and/or statement, or betting that the price of electricity will go up substantially over the life of the equipment.
BTW she paid about $4 a peak watt for the 8 kW system on her house, including batteries, inverter, automatic transfer switch and panel. That was wholesale cost of material only, curiously around the same price for the one on her old house ~15 years ago and her current house ~5 years ago.
AFIK all silicon used in semiconductors is produced from silane, SiH4, which is produced in bulk at "3 nines" (99.9% pure) and then purified to 6 or 7 nines and usually doped before epitaxial deposition or conversion to bulk Si for wafer growth. Per my hazy recollection of a tour of a Praxair semiconductor gasses plant a few years ago.
Per
formatting link
the bulk silane production is:
"Tetrahalosilanes, such as SiCl4, are reduced to SiH4 by hydrides such as LiH, NaH, or LiAlH4."
The MSDS for SiH4 is interesting, for example, if there is a silane leak it will usually ignite immediately on contact with air, but occasionally it will build up and then explode.
But at some point, the cost of your support structure starts to dominate, and low efficiency panels eat you up in the installation costs, esp. when you amortize the costs...
Concentrators can get you more energy to work with, but they also vastly increase the heat, which as most of gurus here now, rapidly age the cells and reduce the lifetime of the panel.
One guy was using thin film concentrators and non-imaging optics to only gather the 'usuable' wavelengths, to get that boost in efficiency without the heat gain, but he still mounted his cells on big water cooled heat sinks!
$750 / $0.15 per KWH = 5,000 KW hours' output for payback.
At 0.75 KWH/day, that's 6,670 days, or 18 years for the panel alone. Add the fixed costs (installation, inverters) and consumables (maintenance, batteries) and it looks even less attractive.
OTOH, 5hrs a day insolation seems a bit light.
Even so, every time I do the extended calculations, the payback period for solar in an urban locale is .... never. Not to mention the environmental impact of periodically recycling all those lead acid batteries.
Wind power pays back very quickly though, if you don't mind the bird strikes.
AFAIK, it starts with white sand -- clean enough for clear glass, about 99 to 99.5% pure -- which is smelted with carbon and sometimes iron to make silicon or ferrosilicon, respectively. This requires an electric arc furnace. (This can also make silicon carbide, but using more sand prevents it, since 2SiC + SiO2 = 3Si + 2CO.) Most goes to metallurgy (all those 2 to
4% Si alloys for transformers don'tcha know..oh and cast iron....), but some goes elsewhere.
You can react silicon with hot chlorine gas (tasty..) and fractionally distill the silicon tetrachloride. This gets it to reasonable purity. The gas can be decomposed on a hot wire, forming chlorine gas and silicon again. Or does it need hydrogen (to reduce it, 2H2 + SiCl4 = 4HCl + Si), I forget. You can run a continuous process where halogens come in and spirit away metal, then deposit it on a hotwire. Since different chlorides decompose at different temperatures, you can precisely control purity.
Afterwards, you get a polycrystalline rod of rather pure silicon, with a thin impurity of say, tungsten in the middle, which can be thin enough inside a thick bar of silicon that it doesn't matter (else you could...bore it out, or, something). This stock is pure enough to melt in a silica bowl and pull crystals from (Czochralski(sp?) method or zone refining).
Speaking of zone refining, you can also start with only somewhat pure silicon and draw a melt zone across it (with fire or resistance or induction heating). I don't know how they do that in a band, through the entire thickness of a rod, without it falling, or if they just heat up a section so impurities *diffuse* along, rather than actual melting. It might be they fit the silicon inside a fused quartz tube, then remove it with hydrofluoric acid to get bare silicon.
Tim
--
Deep Fryer: a very philosophical monk.
Website: http://webpages.charter.net/dawill/tmoranwms
Naaah! It comes from outer space and is brought here by alien space travelers ;-)
...Jim Thompson
-- | James E.Thompson, P.E. | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona Voice:(480)460-2350 | | | E-mail Address at Website Fax:(480)460-2142 | Brass Rat | |
formatting link
| 1962 | I love to cook with wine. Sometimes I even put it in the food.
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.