Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 29
5 Optoelectronics and Photovoltaics Beyond their use as catalysts, critical materials have wide- As was shown in Figure 2-5, the rare earth elements are spread applications in other technologies, including technolo- reasonably abundant; they lie between the 25th and 75th per- gies that could undergo rapid expansion in the future. Many centile in terms of natural abundance, with cerium being the critical elements are used in the display and solid-state lighting most abundant rare earth and lutetium the least. Quoting (SSL) industries. They also are components in many photo- Gschneidner (2011), Shinar noted that rare earth elements voltaic technologies, which are expected to undergo continued are found all over the globe, not just in China, which holds rapid growth. Several speakers noted that research to reduce about 31 percent of the known reserves. In fact, the United the quantities of these materials needed in new technologies States has one of the historically highest-grade deposits in is important, but so is the development of new supplies and the world, located at Mountain Pass, California. improved refining and recycling capabilities. In 1970, China possessed 75 percent of the known rare earth reserves, referring to yttrium plus the lanthanides. At that point, China demonstrated a strong presence in the rare CRITICAL MATERIALS IN OPTOELECTRONICS earth market. However, in the 40 years since then, China’s Rare earth elements are used widely in the field of opto- share of the world reserves of these elements fell to about electronics. Europium is a key element in red phosphors, 30 percent as new deposits were discovered even though mostly in the form of europium-doped yttrium oxide. China grew its reserves through discovery by some 290 per- Erbium-doped fiber amplifiers are critical components in cent (Gschneidner, 2011). fiber optics, and erbium, neodymium, and holmium are Recently, said Shinar, China changed its policies, intro- important dopants in yttrium aluminum garnet (YAG) lasers. ducing production quotas, export quotas, and export taxes; White light-emitting diodes (LEDs) will soon be the pre- enforcing environmental legislation; and refusing to grant dominant user of indium in terms of volume, though ITO new rare earth mining licenses. In addition, China announced will continue to find use in the photovoltaic, display, and it will no longer export rare earths because of rapid growth SSL industries. Chelated heavy metal and rare earth metals, of internal markets and limited reserves, especially the heavy such as palladium, platinum, iridium, and europium, also are rare earth elements gadolinium through lutetium. As a result, used in the display and SSL industries. the price for rare earth materials and products containing rare earth has risen to the level at which mining companies and producers outside of China can make a profit. In 2009, Some Lesser Known Facts About Critical Materials estimated non-Chinese production of rare earth oxides was Though it was noted earlier in the workshop that the 4 kilotons (Gschneidner, 2011). Also, the smuggling of rare rare earth elements are not all that rare, what is not widely earth elements from China appears to be an important source appreciated, said Joseph Shinar, is that there is only a rough of these metals, Shinar said. correlation between abundance and price. If the correlation Another unappreciated fact about the rare earth elements was strong, iridium would be two orders of magnitude more is only about 57 percent of the cerium produced today is expensive than gold, while in fact it is hardly more expensive used. The reason for this is that cerium must first be removed at all. from rare earth flow streams before the other elements can be 29
OCR for page 30
30 THE ROLE OF THE CHEMICAL SCIENCES IN FINDING ALTERNATIVES TO CRITICAL RESOURCES isolated via a countercurrent liquid-liquid extraction process. only in museums, red phosphors are still used extensively As a result, cerium is overproduced by about 50,000 tons per in compact fluorescent lamps (CFLs). And hot on the heels year and must be stored at a real cost (Gschneidner, 2011). As of CFLs are white LEDs for solid-state lighting.” But even expected, the rare earth industry would welcome new large though the demand for europium and yttrium will continue to cerium applications, but so, too, would all other users of rare grow, proven reserves of these metals are outpacing expected earth elements because a balanced cerium market would sig- increases in demand. nificantly reduce the prices of the other rare earth elements. Erbium accounts for about 0.5 percent of mined rare Shinar observed that rare earths can be separated into three earths, and the demand of 700 tons of oxide per year is bal- categories—surplus, balanced, and tight. Only neodymium, anced with supply. Erbium is used as a colorant and as a terbium, and dysprosium fall into the tight-supply category. stabilizer for zirconium in jewelry, but its most critical use Other than cerium and holmium, which is actually one of is in the repeaters used in optical fiber networks that operate the rarest of rare earth metals, supplies and demands for the with a carrier wavelength of 1.5 microns. These repeaters, other rare earths are well balanced. It is also important, he which are optically pumped lasers, are incorporated into fiber added, to remember that the proven reserves of rare earths networks every few kilometers to boost the optical signal. are growing rapidly. Shinar noted the demand for these lasers may slowly wane The solution to the expectation that supplies will become as wireless communications steadily grow. tight as a result of China’s new policies is to increase mining, A niche application for rare earths is in neodymium, Shinar stated. That, in fact, is happening because MolyCorp erbium, and holmium-doped YAG lasers. YAG lasers, which is investing in reopening and expanding its operations at are very robust, efficient solid-state lasers, are staples in opti- Mountain Pass. Workforce training also will be important. cal electronic labs and are used extensively in spectroscopy One lesser-known fact about the critical elements is that the and ultrafast spectroscopy. Holmium-doped YAG lasers supply of ITO, the quintessential transparent conducting mate- also are used extensively by urologists in laser lithotripsy. rial, is totally dependent on zinc production, and the demand Though these are niche applications, they are critical ones for zinc is not expected to grow much in the future because of nonetheless. Of the three rare earths used in these lasers, macroeconomic factors in China, Canada, Korea, and Japan. only neodymium is in tight supply. Though it represents Annual production of indium is approximately 900 tons, with 16 percent of mined rare earth oxides, it is used extensively primary production of 600 tons and recycling and stock- in the neodymium-iron-boron magnets incorporated in wind piles providing the rest. Recycling occurs primarily through turbines. Of the 23 kilotons of neodymium oxides mined, removing the thin ITO layer coated on glass used in LCD 2.8 kilotons go into wind turbines alone. Holmium represents monitors, flat-screen televisions, and other display devices. about 0.1 percent of mined rare earth oxides, and its supply Demand for ITO, which accounts for 85 percent of all of 100 tons of oxide per year exceeds demand. indium demand, is expected to grow by 15 percent per year Indium demand is expected to grow rapidly with the over the next 3 years. Emerging uses are in copper-indium- development of what are called indium group II-V devices gallium-selenium solar cells, electrode-less lamps, mercury for use in the street lighting market. These devices, which alloy replacements, and nuclear reactor control rods. Though are based on new technology for growing high-quality indium prices are much the same as they were 5 years ago, indium-doped gallium nitride, are superior replacements for the price has spiked thanks to China’s recent crackdown on fluorescent and high-pressure sodium lamps in overhanging small lead and zinc refiners amid environmental concerns. street lights. The City of Anchorage, Alaska, estimates that replacing 25 percent of its streetlights with these new lamps will reduce energy costs by 50 percent and save the city some Critical Metals $360,000 per year. In addition, maintenance costs should Shifting his focus from the general to the specific, Shinar drop given that these bulbs have an expected lifetime of spoke next about europium red phosphors, particularly 100,000 hours—over 11 years—compared with 5,000 hours europium-doped yttrium oxide, or yttria. Europium accounts for fluorescent and high-pressure sodium lamps. Perhaps for about 0.3 percent of mined rare earth metals, and the more importantly, the new lamps improve visibility dramati- demand of about 400 tons of oxide per year is balanced by cally because of superior color rendering. supply. Yttrium represents 6 percent of mined rare earths, Indium is also used in ITO photovoltaic devices, displays, and its supply and demand are balanced as well, at about and SSL industries. Shinar focused his remarks on organic 8.5 kilotons of oxide per year. LEDs (OLEDs), which first appeared in a commercial prod- “The main use of red phosphors is as the R in RGB,” said uct in 1998 and is now making major inroads in display Shinar. He explained that ytrium doped with 4 to 6.5 per- devices. In 2007, Sony introduced an 11-inch OLED televi- cent europium produces an intense red phosphorescence sion, but at a cost of $2,500 it was not a big selling item. at 611 nanometers (nm) (Sylvania, 2010). “Even though Since then, at least one company announced it was introduc- RGB monitors and televisions are soon going to be found ing a much bigger OLED television, but it has yet to appear
OCR for page 31
31 OPTOELECTRONICS AND PHOTOVOLTAICS Heavy Metals on the market. OLED displays can have contrast ratios of 1,000,000:1, but LED technology has not been standing Shinar noted that heavy and rare earth metal atom still, and experts continue to argue about which technology chelates, using palladium, platinum, iridium, and europium, produces better displays. are going to continue to be important in the display and SSL One application of OLED with great potential in the industries because they produce the best phosphorescent market is the use of white OLED (WOLED) in lighting emitters. As he explained, “In all OLEDs, injection into panels. Potentially, these devices can be more efficient the emitting layer results in 25 percent singlet excitons and than fluorescent lights and produce more accurate color 75 percent triplet excitons, and in fluorescence only the rendering. In addition, WOLED devices can be flexible and singlet excitons are radiative.” As a result, the maximum can even become mirrors when turned off. Today’s state- internal quantum efficiency is only 25 percent. Heavy metal of-the-art devices produce 87 lumens/watt (lumens/W) at chelate-based phosphors, however, are radiative with triplet 1,000 candelas/m2. The goal is to produce devices that gen- excitons, so their use is critical to achieve high internal erate 300 lumens/W. quantum efficiency for any OLED device. ITO is essential to all of this work because it is the pre- In fact, when researchers from the Universal Display ferred transparent conducting electrode in thin-film photo- Corporation successfully fabricated phosphorescent OLEDs voltaics, and it is the only transparent conducting electrode using heavy metal chelates, they boosted internal quantum in all LCDs and OLEDs. Researchers are making advances efficiencies into the 90 to 100 percent range. This suc- in developing transparent zinc oxide and aluminum oxide cess, said Shinar, explains why this company is now worth electrodes, but those efforts are not yet close to producing a between $500 million and $1 billion. commercially viable product. Shinar noted that iridium, because of its use in these One of the most promising alternatives is poly(3,4- phosphors, should be considered a critical metal. It is the ethylenedioxythiophene): poly(styrenesulfanate) (PEDOT:PSS). least abundant of the platinum-group elements, yet today it is Though this material has been studied for over 10 years, the priced lower than gold. Major commercial sources of iridium recent development of a fabrication process that creates are found in South Africa, Russia, and Canada. Iridium is multilayered PEDOT:PSS devices has produced a break- difficult to refine and is produced in small quantities, but through in organic photovoltaic and OLED performance supplies have increased in response to a four-fold increase (Kim et al., 2011). The new process involves blending these in demand to 334,000 ounces in 2010, largely a result of polymers with polyethylene glycol (PEG), immersing the the inclusion of iridium crucibles in backlit LED televisions blend in PEG, and then annealing. Even though transmis- and the increased demand for iridium-tipped automobile sion goes down with each additional layer, sheet resistance sparkplugs. “The sharp increase in demand and the small, also goes down, and that, said Shinar, is important. These relatively illiquid market for iridium had a significant impact PEDOT:PSS sheets are smoother than ITO, which is good on price,” said Shinar. In August 2011, iridium was priced at for OLEDs, and their refractive index is lower, which reduces $1,050 per ounce (eBullionGuide.com, 2011). internal reflection and increases light output. Efforts to develop room-temperature phosphors free of In fact, said Shinar, multilayer PEDOT:PSS OLEDs are heavy metals have begun, but the best results so far still fall up to twice as efficient as a standard ITO OLED. And in short of the mark. Shinar wondered if more research in this more recent work, which his group has not yet published, direction should be initiated. One possibility would be to a two-layer PEDOT:PSS OLED produced a maximal lumi- exploit triplet-triplet annihilation to produce singlet excitons nous power efficiency of 100 lumens/W without coupling that could increase the internal quantum efficiency well enhancement tricks. “With microlens arrays, which typically beyond 25 percent. can double the out-coupling enhancement, these devices In closing, Shinar noted that “for optoelectronics, the would be beyond 200 lumens/W,” said Shinar. critical in critical resources is questionable. There is no It is important to remember, though, that ITO devices are single silver bullet because the situation has improved, and a moving target with continually improving performance. instead there are many potential silver bullets for different Recently, for example, chlorinated ITO-based OLEDs problems.” showed impressive efficiencies (Helander et al., 2011). The power efficiency reported for these devices exceeded KEY MINERALS IN PHOTOVOLTAICS 200 lumens/W, which is approaching the Department of Energy’s goal of 300 lumens/W. Not too long ago, this was One reason that PV technology is such an exciting area considered a pipe dream, said Shinar. The external quantum today, said Ken Zweibel, is that solar energy is such an abun- efficiency for chlorinated ITO is “an amazing 53 percent,” dant resource (Figure 5-1). The sheer size of the solar energy he added. “For every two electrons you inject into the OLED “reserve” dwarfs all other potential sources of renewable you get one photon out, and not just out, but in the direction energy and is more than an order of magnitude larger than all you want.” coal, uranium, petroleum, and natural gas reserves combined.
OCR for page 32
32 THE ROLE OF THE CHEMICAL SCIENCES IN FINDING ALTERNATIVES TO CRITICAL RESOURCES finite renewable 25-70 WIND per year 215 Waves1 0.2-2 total SOLAR 10 Natural Gas 23,000 TWy/year 3 -11 per year 240 OTEC total Petroleum 2009 World energy 2 – 6 per year consumption Biomass 16 TWy/year 3 – 4 per year 90-300 HYDRO Total 0.3 – 2 per year TIDES Geothermal 0.3 per year 2050: 28 TWy Uranium 900 Total reserve © R. Perez et al. COAL FIGURE 5-1 Solar energy dwarfs all other sources of renewable and finite energy sources. SOURCE: Perez (2009); Perez et al. (2011). Furthermore, an important characteristic of solar energy has Over 35 years, each doubling in sales volume has produced been its historical reduction in price, Zweibel explained. a 20 percent drop in price. Today, said Zweibel, “prices are Figure 5-1 like a stone, with the cost now down to $1.30 or A plot of the price of the global average solar energy power dropping taken from sourceper watt for the modules, whereas only a few years $1.40 (Perez) module in constant dollars versus cumulative sales shows a consistent relationship from 1976 to 2009 (Figure 5-2). vector editable closer to $3 per watt.” ago, it was At the same time, worldwide photovoltaic module pro- duction has risen exponentially over the past decade, with China’s entry into the market providing a huge boost in world output. In large part because of increased production from China, output nearly doubled in 2010. China first entered the market in 2006 and now accounts for over half of the world’s production of PV modules. Chinese production has grown so quickly that prices worldwide have plummeted. As a result, margins are now very thin, and many PV module manufacturers, including those in China, are facing financial difficulties. The beneficiaries of the plummeting cost of PV modules are those who install PV systems, and as production has increased so has the number of PV installations (Figure 5-3). In 2010 alone, the total peak megawatts installed more than doubled, with Germany leading this dramatic uptake in PV use. In 2011, growth in the amount of PV installed returned to FIGURE 5-2 Between 1976 and 2009, the price of solar energy more normal levels, which accounts for the glut in supplies. power modules has declined as cumulative sales increased as a In terms of cumulative installations, PV now produces result of aggressive pricing for market share. about 40 gigawatts (GW) of power worldwide. Growth in SOURCE: Paula Mints, Principal Analyst, Navigant Solar Services output is similar to that seen decades earlier when natural gas Program, March 8, 2010. R02178 Figure 5-2
OCR for page 33
33 OPTOELECTRONICS AND PHOTOVOLTAICS FIGURE 5-3 Production of photovoltaics has exploded over the past decade. (Consider U.S. solar cell (PV) and module manufacturing market share.) SOURCE: PV News and Navigating Consulting, http://energy.gov/articles/competition-worth-winning [accessed December 22, 2011]. and nuclear power started becoming significant components but less competitive are amorphous and thin-film silicon on of the energy infrastructure. glass. Emerging technologies that are less assured of success Germany, said Zweibel, is a special case and is far ahead include organic dye-based “Graetzel” cells and plastic cells. of every other country in the world in terms of PV uptake. The most important materials that are needed to produce PV now supplies about 10 percent of peak energy use in the solar power today include silicon, silver, tellurium, and middle of the day, or about 3 percent of the country’s total cadmium, with steel, aluminum, and copper being used in the energy usage. Using comparable measures, wind provides contacts and housings and other bulk components. Emerging about twice the total amount of power as solar does in technologies could require significant quantities of indium, Germany. selenium, molybdenum, gallium, germanium, arsenic, and Photovoltaics are semiconductors operating at about ruthenium. Many other metals, such as nickel, zinc, and tin, 10 percent efficiency. It takes 10 km2 of PV to produce are used in minor amounts. 1 GW of power. For thinner PV modules, that translates into There are a number of bottlenecks. For example, the risk 10 cubic meters of crystalline silicon per gigawatt of power. and timing of investment, including the unpredictability and The Department of Energy’s Solar Vision calls for PV to rapid alteration in demand, have led to silicon shortages. account for 10 percent of U.S. electricity output by 2030, “Because of the capital cost for making purified silicon, which would be approximately 600 GW, compared to today’s there was a time when the silicon purification industry was output of around 20 GW. The International Energy Agency out of sync with the demand from solar,” said Zweibel. The predicts that solar will generate 50 percent of the world’s demand for silicon by the solar industry is now larger than electricity by 2050, though Zweibel found this figure hard the demand from all the other semiconductors in the world. to accept. “That transition was hard for traditional silicon purification companies to grasp.” Other bottlenecks involve extraction of the elements Meeting Future Demand needed to meet growing demands from the solar industry. Zweibel listed a number of key technologies that will If there is not enough of a particular element available to play a major role in the future growth of PV as an electricity meet the immediate or short-term future demand, it may not source. Crystalline silicon currently holds about 90 percent be economically feasible to increase extraction in a timely of the market, with cadmium telluride accounting for the manner. Similarly, there may not be refining capability to remaining 10 percent. Emerging and promising materials meet increased demand, or it may not be economical to refine include copper indium selenide alloys with gallium and ores with low concentrations of the needed element. Finally, sulfur, and type III-V multi-junction semiconductors for supplies may exist, but they may not be accessible for pur- concentrator-containing solar cells. Commercially viable chase because of market or political forces.
OCR for page 34
34 THE ROLE OF THE CHEMICAL SCIENCES IN FINDING ALTERNATIVES TO CRITICAL RESOURCES Specific materials have unique issues. The availability per year, molybdenum 18,000 metric tons per year, and of coated glass substrates has been a problem for short selenium 9,600 tons per year. periods of time. Materials such as tellurium used in cadmium Research may yield solutions to these impending supply telluride, indium used in copper indium selenide alloys, and problems, Zweibel said. If not, the most important recent gallium could have price and supply issues at mid- and long- commercial PV technology—the development of cadmium term time frames. Of these, indium is also used in significant telluride—will have an uncertain future because of tellurium amounts in liquid crystal displays. availability. In addition, the most important emerging com- Today, materials such as tellurium, indium, gallium, mercial technology—copper indium gallium selenide—will molybdenum, and selenium add between $0.002/W to have an uncertain future because of indium, selenium, and $0.03/W to PV costs, with silver adding $0.09/W, numbers gallium availability. It is essential, he said, to find new sup- that Zweibel characterized as being small. However, the costs plies and develop new refining capabilities for tellurium and of each of these materials can rise dramatically in response indium. Otherwise, all scenarios to meet demand for PV to even small changes in supply and demand. Given that will be highly challenging. “The lack of U.S. commitment $0.10/W to $0.20/W separates the best PV modules from the to extraction and refining inside our borders is a concern for least competitive modules, changes as small as $0.03/W can U.S. PV competitiveness,” Zweibel concluded. be significant. “Something that goes from 3 cents to 15 cents per watt could be prohibitive,” said Zweibel. DISCUSSION In response to a question about the role of chemistry in Using Less Material developing new materials, Shinar said that chemistry lies at There are various strategies for reducing the amount of the heart of developing WOLEDs and OLEDs, since each material needed to produce electricity. One approach is to use of these devices depends on complex organic chemistry and thinner layers of material, though that reduces the amount of polymer chemistry. As an example, he said that chemists light absorbed. That problem can be solved by adding back- have been studying polyanilines as transparent conducting side mirrors, which can enable the thickness of the semi- organics and are trying to solve stability issues. The problem conductor layer to be reduced in some cases by a factor of 10. is not that chemists do not want to work on developing these Boosting device efficiency reduces the amount of material materials but that there is not much funding available to do needed. Doubling the efficiency of a device means using half so. In terms of inorganic materials, research is proceeding of the material to generate the same amount of electricity. slowly, and these materials are hard to develop. Recycling, both internally in terms of manufacturing waste In response to a question about potential environmental and externally as far as recovering materials from old mod- issues associated with the widespread use of cadmium, ules, can have a significant impact on supplies. Zweibel said that, because the manufacturers take the mod- Already, such strategies have played a role in reducing ules back at the end of their lifespan and recycle all of the material demands from PV manufacturing. For example, the cadmium telluride, the system is actually a closed loop. Also, cadmium telluride layer has been decreased from 3 microns cadmium telluride is much less dangerous than cadmium. to 0.67 microns with little difficulty in light absorption. Thin- For those reasons, there has been no backlash against the use ner copper indium selenide alloys—0.75 microns compared of very thin, very stable layers of cadmium telluride in PV. to 2.0 microns—have reduced demand for indium, selenide, Zweibel replied to a question about the impact of the and gallium. Silver replacements are now being tested, and dramatically lower price for silicon cells on the technol- Zweibel predicted that silver will be replaced eventually in ogy development by noting that copper indium selenide PV modules. technology is already suffering a good deal of push-back By 2030, such efforts could reduce the per-watt demand for because it has to make price goals that have moved down so rare metals. Zweibel predicted that tellurium use could drop quickly. He added that, although he once thought that thin from 100 to 16 metric tons/GW. Indium usage could drop from films would dominate the future of PV, “I no longer think 30 to 9.4 metric tons/GW. Gallium usage could fall from 8 to that, and I would say that it is going to be a horserace for 2.3 metric tons/GW. the next 20 years.” Although such reductions are important, demand will In response to a question about mining bismuth telluride nonetheless soar if the world truly meets the goal of get- as a significant source of tellurium, Zweibel noted that there ting 10 percent of its electricity from PV. Reaching this are places in the world where this mineral is accessible goal implies 600 GW/year annual production, which would and where the tellurium is highly concentrated, as high as require 10,000 metric tons of telluium per year, compared to 17 percent in at least one instance. These deposits are likely 120 metric tons used today. The total availability of tellurium to change the supply issue dramatically when the current pro- is about 2,000 metric tons per year today. Indium use would duction sources—copper, zinc, lead, and gold mining—start reach 6,000 metric tons per year, gallium 1,500 metric tons proving inadequate. He noted, too, that tellurium is the most
OCR for page 35
35 OPTOELECTRONICS AND PHOTOVOLTAICS abundant metal in the universe with an atomic weight over of $2.50/W to $3.50/W, with the module’s cost running at 40 Daltons, even though it is depleted in the Earth’s crust $1.00/W to $1.50/W. For residential applications, installation (Cohen, 1984). There are large deposits in undersea ridges, costs are higher. The Germans have worked out the most effi- and some companies are starting to show interest in undersea cient installation methods, and their costs run about $1.00/W mining at these ridges, which are rich in many other materials to $1.25/W with total system costs at $3.50/W. Residential as well (Hein et al., 2001). installation costs also run higher because of the soft costs In response to a question about the role of chemistry in involved, such as marketing and sales. He noted that the developing new materials, Zweibel said that chemists play a United States can learn much from the German experience. critical role in two areas. In the development of mainstream Finally, when he was asked about the role of chemists PVs, the most effective researchers in this area are chemists. in the development of photovoltaics, Zweibel said that they It is chemists who develop the understanding of how mate- are at the center of that process, despite the substantial rials behave during processing and about the fundamental involvement of electrical engineers, physicists, and others. behavior of the materials themselves. The second area where It is “important to understand how that chemistry happens chemistry is important is in manufacturing, which is really both in terms of the processing and in terms of the nature a chemical engineering problem and relies heavily on the of the material itself.” Chemists also are needed in the knowledge of chemical engineers. large-scale manufacturing of photovoltaics, which is very When asked about installation costs, Zweibel said that dependent on chemical engineering. “Chemists are really large-scale installations are largely mechanized and that the the heart and soul, to a great degree, of this technology and cost per watt can be as low as $0.20/W out of a total cost of photovoltaics.”
OCR for page 36