Analysis of Solar Energy - Can it supply human energy needs? -----------------Warren D. Smith Mar 2001------------------- AVAILABLE SOLAR TECHNOLOGIES: At present there are 3 competing kinds of photovoltaic modules commercially available (single crystal Si, poly-Si, and amorphous hydrogenated Si) plus mirror concentrator arrays, plus glass "greenhouse hot box" technology (currently mainly used to produce hot water, not any "portable" form of energy such as electricity) a fourth and fifth technology. Ideas already incorporated into current commercial photovoltaic technology are: * micro-patterned surfaces to increase light absorption (BP solar), * anti-reflection coatings, * capability of absorbing light from both front and back surface (Photovolt), * use of tempered glass for higher strength, * and use of low-iron hi-transmission glass, * indium/tin oxide conductive transparent electrodes, * split spectrum multijunction stacked thin film cells deposited on flexible stainless steel sheet (unrolled from rolls and rolled onto rolls) by chemical vapor deposition - the topmost junction layer has a high bandgap energy and absorbs blue photons, the lowest layer has a lower bandgap energy and absorbs red ones (Ovonics/Unisolar), * encapsulation in plastic and/or glass, * built-in protection diodes, * both flexible sheets and rigid sheets, * use of wafers "recycled" from semiconductor industry discards to save money (AstroPower Inc.) and soon-to-come technology (scheduled for summer 2001: AstroPower "APex") will include * poly-Si continuous ribbon extruder without need for sawn ingots and consequent sawdust waste and expense. POSSIBLE ALTERNATIVE TECHNOLOGIES (THAT DON'T LOOK COMPETITIVE): (I) It has also been suggested that photovoltaic cells could be constructed based on thin films of organic semiconductors rather than silicon-based semiconductors. But the best efficiency so far achieved by such a solar cell is only 2.5% [S.E.Shaheen & 5 others: 2.5% efficient organic plastic solar cells. Applied Physics Letters 78,6 (2001) 841-843] as opposed to the records for amorphous hydrogenated Si (13%, Ovonics/Unisolar), and poly-Si and single crystal Si (20-30%). (Incidentally, efficiencies for complete weatherproof installable commercial solar MODULES are necessarily well below record efficiencies achieved in the lab for single solar CELLS - do not be confused. This is due to light loss necessitated by protective packaging, area loss due to imperfect coverage by tiled solar cells, and the use of average cells rather than selected best cells.) The hope expressed by proponents of organic semiconductors is that they will be ultra-cheap since the required materials will be capable of being "printed on" surfaces as "inks" - dissolved in solutes which then will evaporate. (Actually newsprint ink is based on non-evaporating oils which instead solidify due to air oxidation. But perhaps printing of molten organics could be done?) Organic semiconductor LED displays are commercially available and are stable enough for use in automobiles. Meanwhile Ovonics/Unisolar's thin amorphous Si:H films are deposited onto stainless steel by plasma enhanced chemical vapor deposition and other required films (e.g. tin oxide) are deposited in vacuum chambers or sputtering systems - requiring more sophisticated manufacturing technology. This all sounds good... but unfortunately, in fact, organic semiconductor devices so far HAVE been made using vacuum chambers, tin oxide deposition, etc, and have NOT been made by the hypothetical cheap printing techniques, and they currently are MORE, not less, expensive than inorganic semiconductor devices. (Also, let's face it, silicon is a more common element on Earth, than Carbon is. We in fact are running out of carbon - that is the problem.) So the hopes of the proponents of this technology currently are just that - mere hopes. It is certainly conceivable to me that other thin film materials, perhaps amorphous materials, could be better than Si, such as SiC, C, or BN. Indeed see Applied Phys Lett. 61 (Dec 1992) 2805 and 59 (Jul 1991) 69 for amorphous carbon, whose bandgap and structure seem in principle adjustable between graphite (zero bandgap) and diamond (large bandgap). Amorphous semiconductor cells are very little understood theoretically, at least by me, so serendipitous discovery of good materials of this kind would seem possible. It has also been suggested that the use of such complex crystal semiconductors as GaAs could lead to greater efficiencies than Si. However, GaAs decomposes after many years exposure to air into, e.g., highly toxic arsenic and gallium oxides, and covering vast areas of the Earth's surface with such materials is presumably environmentally unacceptable. Similar remarks would seem to apply for cadmium-based semiconductors. Si is the only semiconductor material with a decent bandgap-match to the sunlight spectrum, the capability to fabricate good quality crystals, good material cheapness, and non-toxicity. (II) "Bio-mass" is another idea for using solar energy. However, the "efficiency" with which land plants (the best of which seem to be fast-growing grasses) convert incoming solar energy per acre into useful bio-mass is apparently very low, usually under 3% even under favorable conditions, (although I've seen an unsupported claim sugar cane achieves 8%), and in practice even less since some parts of the plants, such as roots, often cannot be harvested. Indeed, table 25.5 of [Anne Fege: Energy from biomass, chap. 25 of Solar Energy Handbook,McGraw 1981] gives data on energy and mass yields per acre per year for 23 different kinds of plants, including sugar cane. The largest energy yield among land plants was Exotic Forage Sorghum in Puerto Rico, yielding 68.7 tonnes/hectare/year, i.e. (assuming 7500 Btu/lb) 1.21*10^9 joules/hectare/year. This is only about 1.5% of the incoming solar energy on that hectare per year! Even higher yielding than this Sorghum is cultured algae in California yielding 1.50*10^9 joules/hectare/year, but that still only represents 1.9% efficiency. Furthermore, burning these materials is also low efficiency since they are poor fuels (high water content) compared to, e.g. coal. On the other hand, many of these plant materials would otherwise be wasted, energy-wise, and we get them "for free" as a byproduct of other agriculture. (Even then, the yield is less than one might imagine since responsible farmers leave much of the plant in the ground to prevent soil erosion.) Also, garbage, sewage, etc, have been suggested as fuels for the same reason (possibly after bio-conversion to methane), but their burning also can be low efficiency and sometimes environmentally damaging. It might be that selected algae or bacteria would generate bio-mass a lot more efficiently than land plants from incoming solar energy per acre. Theoretically the chemical mechanism of photosynthesis could achieve efficiencies of up to 13%, but the best achieved values in outdoor experiments are below 5%. Microbes can also generate hydrogen gas directly from sunlight and water. Reported "efficiencies" (hydrogen combustion energy / light input energy) have ranged from 0.2% for cyanobacteria to as high as 6-8% using Rhodobacter [Miyake, J. and Kawamura, S.: Int'l. J. Hydrogen Energy 39 (1987) 147-149] which gets its hydrogen from organic acids, not water, producing also CO2. This of course is "cheating" but in combination with CO2-fixing photo-organisms and anaerobic organisms for converting hydrocarbons to organic acids we'd have a fully biological system whose net effect would be to input water and light and output H2 gas. Photosynthetic microbes and plants can also produce oil-like substances. So anyway, it seems that bio-mass is in net, not as efficient as photovoltaic cells for generating either electricity or hydrogen gas, nor is it as robust to, e.g., temperature variations and nor is it as able to operate in, e.g., arid deserts where there is the most sunlight (but no water) - and all that probably would remain true even if maximum theoretical chemical efficiencies were somehow achieved by future lifeforms. Biomass may however be competitive in efficiency if the goal is to generate oils. [Renewable biological systems for alternative sustainable energy production (FAO Agricultural Services Bulletin - 128) 1987 Edited by Kazuhisa Miyamoto Osaka University Osaka, Japan FAO = Food and Agriculture Organization of the United Nations ISBN 92-5-104059-1]. (III) The "greenhouse hot box" idea, used in many "solar water heaters", involves an insulated box, whose interior is black, and whose top side is covered by 1 or 2 sheets of glass. (An idea so far unused is to replace the sheets of glass by a hydrophobic silica-based "aerogel" a highly insulating, transparent, and low-density material, but which unfortunately currently costs about $100 for a square-foot sheet.) Sunlight, either direct-ray or diffuse, shines into the box, causing it to become hot. The heat cannot get out due to the insulation and because the glass is opaque to infra-red. Thus we are taking advantage of the fact that sunlight is NOT a terrestrial-temperature blackbody spectrum. One can take even more advantage of this effect by using low-iron glass with enhanced transmission of visible sunlight, evacuation of air, anti-reflection coatings, and also by using special coatings on that glass intended to reflect IR; also the black surface can be a "wavelength selective" material intended to have high absorption in the range .3-1.9 microns but low absorption (and low emissivity) in the IR range 5-15 microns. Black nickel oxide on aluminum is an example of such a material with an emissivity ratio of about 15. The heat from the hot box may be extracted by pumping water through coils whenever the box is substantially warmer than the hot water tank. Also, heat storage in other thermal reservoirs, besides just a water tank, has also been done - for example pebble beds and tanks containing cheap melted salts with high heat of phase change: Na2SO4.10(H2O) ("Glauber's salt," or sodium sulfate decahydrate, melting 37.8C) and CaCl2.6(H2O) have both been used (in combination with nucleators to prevent supercooling). A configuration known as the "thermal siphon" in which hot water rises out of the top of the solar panel, even allows avoiding the need for a water pump and temperature sensor. Problems with this kind of device have included freezing, leakage, corrosion by the hot water, and pump and valve failures. The solar hot water heater is excellent for supplying hot water, achieving efficiencies of over 50%. (And it has been estimated that about 4% of USA energy consumption goes to heating water for home or industrial uses.) But this technology seems less suited for the purpose of supplying world energy needs more generally. If heat engines were developed that worked with small temperature differences (i.e. <100C) then one could generate electric power, but even with maximum possible thermodynamic efficiency - meeting the Carnot bound - a heat engine could not garner more than 25% efficiency from a 400K-300K temperature difference. Thus the net efficiency of the entire system (heat engine and solar collector) would be 12% at best. This efficiency would be competitive with photovoltaic panels, but would involve corrosion, freezing, and leakage worries, and the need for moving parts, unlike photovoltaic panels. The prices at present seem to be comparable for hot-box/water collectors than and photovolatic panels, per square meter. So I think the best way to use hot box collectors is either (1) purely for providing hot water or (2) to make combination photovoltaic AND hot box collectors in which waste heat from photovoltaic cells is used by the hot box, and then used to generate additional electric power. Such combination collectors are not presently commercially available, to my knowledge, but they would presumably yield a substantial energy bonus from photovoltaic panels while (if used on a large scale) for which one would pay a smaller cost increase factor than the energy increase. ANALYSIS OF WORLD ENERGY NEEDS: Current world energy consumption (year 2000) is about 12600 GW-years per year, i.e. 2100 watts per capita times 6 billion people; USA use is about 5 times larger per capita than world average, i.e. 10.5 kwatt. [UN figures.] Solar COULD supply this if about 400,000 square kilometers were covered with solar panels (assuming 126 watt/meter^2 PEAK power production from solar modules, i.e. about 12.6% efficiency, which corresponds to the most efficient commercial modules available in the year 2001 for Earth use, and this corresponds to about 31.5 watt/meter^2 AVERAGE production over the day/night cycle - i.e. assuming clear skies always and assuming storage/retrieval and power conversion and transmission losses may be neglected. Those assumptions may be very optimistic - but I'm confident not overoptimistic by more than factor of 5.) Storage and Retrieval of energy would be a necessity since you don't get solar power at night - although the possible hypothetical future development of lossless high capacity superconductor transmission lines would eliminate the need for energy storage since the sun is always shining in some time zone. Unfortunately present battery technologies are neither cheap nor reliable enough and/or are too toxic and environmentally impacting (e.g. lead batteries). Also technologies other than Si (Cadmium, Gallium-Arsenide) seem too expensive and/or toxic and environmentally impacting to be considered. Hydroelectric water pump-up/falldown and electrolysis of water to produce H2 gas which could be stockpiled in natural or artificial caverns, then burned in turbogenerators, are better large scale storage/retrieval technologies than batteries. But they suffer a loss factor of about 3. The hydroelectric idea may ultimately be inadequate since there simply is not enough hydroelectric power in the world. The H2 idea could work in conjunction with present day power plants or in conjunction with turbogenerators used during daylight with solar concentrators. (Commercial water-electrolysis technology is presently highly advanced and reliable - thanks to nuclear submarines.) I believe a far better idea would be to use a giant coil of superconductor to store magnetic field energy. This idea seems to me to scale up superbly since the cost grows linearly in the size, but the energy storage grows cubically. No other energy storage method has that superlinear-growth property. Imagine a 20-mile diameter superconducting circle in some unpopulated area (Antartica?) generating a 5-Tesla magnetic field. Energy storage would be of order 10^20 joules, i.e. 3000 GW-years. So about 4 such devices could store 1 full year's worth of world energy consumption! HOW MUCH AREA IS 400,000 (km)^2? HOW HARD IS IT TO COVER THAT MUCH AREA? OK, 400,000 (km)^2 is "only" about 1/372 of the land surface area of the earth. Sounds small? Well, it is about the same as the combined total land (non-water) area of Great Britain plus Japan. It is also about 1.3 times the area of Arizona. THOSE don't sound so small. But hey, this area is only about 1/22 of the Sahara Desert. No problem? The total area of all roads in the world in 1982 was - assuming 20 giga km of roads [Brittanica Yearbook 1982 page 351] on average 6 meters wide: 120,000 (km)^2. Thus we are talking about a solar collector area 3-4 times larger than the area of all world roads in 1982. That is a lot of building to do. On a per capita basis, this is about 67 meter^2 per capita, i.e. about an 7x10 meter plot per capita. (USA citizens should, for their "fair share", use 335 meter^2, i.e. a 5x67 meter plot, or about 1/12 of a football field, i.e. 1/12 of an acre, roughly, per US capita). At the cheapest current (year 2001) solar module prices $4.25 per PeakWatt, that would cost US citizens $178,500 per capita to buy their fair share of the Solar modules. Is building all this feasible? Well, assume it all has to be built in 20 years (or that it only lasts 20 years, hence has to be continually rebuilt every 20 years forever! - the maximum full warranty currently now offered by a solar module manufacturer is 20 years, plus Siemens Solar offers a limited 25-year prorated warranty). In that case you have to build 20,000 (km)^2 per year, i.e. about 55 (km)^2 per day. (So much for that cliche "Rome can't be built in a day"!) WHAT ABOUT ENERGY REQUIRED TO BUILD IT? A Siemens study has concluded that today's solar modules have an "energy payback time" of about 3 years, i.e., solar modules really do provide positive amounts of energy over their lifetimes of >20 years. That is reassuring, although maybe not tremendously so. WHAT ABOUT PRODUCTION LEVELS? The annual world production of PAINT [UN industrial commodity stats yearbook 1998] is about 2.7 million tonnes per year, which is enough to paint about 30000 (km)^2 of wall area per year (standard housepainting estimates). So, we have to build solar receiver surface area at a rate comparable to the total rate at which humanity now can paint ALL surface area! The total world production of rectangular SHEET GLASS [also UN statistics] was about 757 (km)^2 in 1998, so we have to build solar collector area at about 26 times the total rate at which the world currently manufactures plate glass area! The total world production of ALUMINUM cans/packaging was about 2.5 million tonnes in year 2000, which pretending it all was aluminum soda cans (each weighs 1/33 lb), would be about 1.8*10^11 soda cans per year, i.e. 30 per capita with 6*10^9 world population. That many soda cans adds up to about 5,900 (km)^2 of Al-sheet-metal area. So, we have to build solar collector area at a rate about 3-4 times the rate at which the world has been producing Aluminum for packaging (which I am told is a significant fraction of world Al consumption). The New York Times produces about 2100 km^2 of PRINTED PAGE surface per year (counting only 1 side of the page as "surface") so we are talking about a total solar collector production and installation rate equivalent to the combined rate at which about 10 major newspapers produce surface area. That is a hell of a manufacturing rate (indeed, high-volume printing is widely regarded as the fastest manufacturing technology there is) compared to all previous rates for electronics. WHAT ABOUT SILICON? The total world production of 99.9%-pure poly-silicon is about 3200 kilo tonnes per year (BUT "semiconductor grade" silicon suitable for solar cells is far purer, and perhaps 40 times more expensive, than mere 99.9% pure feedstock Si! Only a few percent of world silicon production is used by the semiconductor industry at present and hence hyperpurified) which, assuming we'd need 100-micron thick Si to cover those 20,000 (km)^2 per year (100um is the minimum thickness for good light absorption by single crystal Si, and this is a good deal thinner than current solar cell Si wafers)... would be INSUFFICIENT by a factor of about 20, and is insufficient by a factor of about 700 if we are talking about semiconductor grade Si. OOPS!! OUCH!! That really hurts. Ovonic/Unisolar deposits amorphous Si:H and other thin films (only totalling 1 MICRON thick - which is possible since this material absorbs light much better than crystalline Si!) on a stainless steel backing sheet .125mm thick. Thus their technology uses at least 100 times less silicon. (It is also cheaper per square meter. But, their cells are less efficient than crystalline Si based cells, so they need about TWICE as many square meters, which, so far, ends up costing them about the same amount, or slightly more, per PeakWatt.) The whole Ovonic/Unisolar thin-film philosophy seems very praiseworthy as far as resource consumption for huge solar collector areas is concerned. But even so... consider the fact that one of their thin film components is the transparent conductor tin oxide (or perhaps indium/tin oxide). The total amount of tin in a tin oxide film .25 micron thick and 800,000 (km)^2 in area is about 400,000 metric tons, which may be compared with the world total tin production [CRB commodity yearbook] of 206,000 metric tons in 1998. Still over the 20 year build-time, this is only 20,000 metric tons per year, i.e. 1/10 of the annual tin production, not a severe obstacle. But if the tin oxide contains a substantial amount of indium, then this WOULD be a severe obstacle since the annual production of indium is only about 240 metric tons, 80 times too little. (Perhaps Indium/Tin oxide could be replaced by doped Zinc Oxide, which has the advantage that zinc is considerably less costly and more common than tin.) Also, if their amorphous silicon contains a substantial amount of the rare element Germanium (as has sometimes been suggested; elements that have been considered for mixing in with the amorphous-Si include Ge, F, H, C, Cd, Te, B, P, Sn, Cl), that similarly would be a major problem, since only about 80 metric tons of Germanium are produced worldwide each year. Stainless steel must contain at least 10.5% CHROMIUM, so to cover 40,000 (km)^2 per year with this much stainless steel would consume 1.6*10^11 kg of chromium per year. That is 33% of the total world production of Chromium in 1998 [UN Industrial commodity statistics yearbook 1998] so it too does not seem to be a limitation. The total 800,000 (km)^2 would only consume 6.5 years worth of 1998's chromium production. So... while all this might be possible maybe, we can see it certainly would be the most tremendous engineering feat the world has ever seen and would require production ramp-ups on fundamental commodities by multipliers which seem just about impossible to achieve in the time we have left. AND the above was under optimistic technical assumptions and assuming no further population growth and no further growth in power consumption per capita; fat chance of that... WHAT ABOUT COST? At present, photovoltaic solar collectors cost about $4-$10 per PeakWatt, i.e. about $16-$40 per average watt, of generating capacity. And this does not count installation costs, power convertor costs, real estate costs, and storage/retrieval costs. Meanwhile conventional power plants cost $1-$2 per watt of generating capacity - mostly closer to $1. So, solar generation is in general un-economical at present (year 2001) by a large factor - certainly at least 10. If however the cost of solar collector area can ever be brought down to about the cost of 1 sheet of tempered glass ($38/meter^2 in 2001 from Peninsula Glass Co.; acrylic sheet has comparable costs to glass) alone - which would seem to be the cost lower limit, since "float" sheet glass production is a very mature field! - while still maintaining 100 PeakWatt/meter^2 output, then solar might indeed start to become competitive in terms of capital cost. (We assume also that the real-estate costs are neglectable...) So the challenge is to approach this lower cost limit set by the cost of 1 sheet of glass. Light concentrating mirror arrays sacrifice 40-60% immediately because they can only use direct-beam sunlight not diffuse light. But presumably mirrors can be built more cheaply than photovoltaics per meter^2. And mirrors can use turbo-generator technology which is typically 30-40% efficient, considerably more efficient than photovoltaic cells, so the efficiency contest comes out about even. Indeed numerous demo plants based on mirror arrays and turbogenerators have been built in the .1 to 10 MW peak capacity range in sunny arid areas of Spain, France, USA, Russia, and they tended to achieve outputs of 80-125 PeakWatt/meter^2 of mirror area, which indeed is comparable to the PeakWatt/meter^2 numbers from available photovoltaic module panels. (Mirror arrays are superior in sunny high-altitude locations where direct sunlight is a larger fraction of insolation versus scattered sunlight. The opposite is true in low-altitude cloudy areas.) But: these mirrors need to be gimbal-mounted with sun tracking, adding to the cost. If the gimbal-mount and sun tracking costs could be made negligible compared to the sheet glass cost, then solar mirror arrays would indeed be able to approach our cost limit... except for the need for conventional turbogenerator equipment at the foci, which presumably would cost about as much per watt-capacity as they always do, and must be designed for PeakWatt capacity 4 times average capacity, thus forcing un-economicalness by a factor of at least 5 in capital costs. Inherently. Oops. CONCLUSIONS: It seems plausible that, no matter what, solar is ALWAYS going to cost AT LEAST 5 times more than conventional power plants, in terms of CAPITAL COSTS, and it may cost up to 40 times as much. The MAINTENANCE costs are presumably also going to be larger by comparable factors. On the bright side - zero FUEL costs for solar. The world PRODUCTION LEVELS needed to build and maintain the needed solar collectors would need to be increased above 2001 AD levels for some or all of the following commodities by the following factors: by a factor of 26-130 for sheet glass, 3-20 for aluminum, 100 for indium, 350 for Germanium, 20-100 for silicon 99.9% pure feedstock (plus this would then also have to be hyperpurified if used to make photovoltaic Si arrays, requiring an increase in world production of hyperpure Si by a factor of 600-5000; fortunately amorphous Si:H arrays would need 50 times less hyperpure Si since thin films suffice, so that the above numbers could be divided by 50. But these are about 2 times less efficient and hence would need twice the land area, and they tend to decay in efficiency by 20% with time.) 4 for turbogenerators, and 100-500 for land surface area consumption versus area of all roads. Furthermore during the initial years of solar construction, the amount of energy used during the construction process would exceed the amount it produced (energy payback time >=3 years), causing considerable energy demand in a world which only was doing this because it was running out of energy. (But the required consumption of Chromium and Tin would be only 33%, or less, of current annual production levels.) The 2001 AD rate of annual world production of solar module generating capacity is inadequate by a factor of at least 1000 and the costs of those modules are uncompetitive by a factor of 10-100. Energy storage/retrieval systems and/or superconducting transmission lines would be needed which are not currently available (and the latter may be, in fact, impossible). Research into high-Tc superconductors should be continued.