Many life-cycle analyses have been made of both silicon and thin-film solar cells. In 2007, Raugei et al. [10] published a careful study of the environmental effects of both silicon and thin-film solar cells using actual production data from Europe. For polycrystalline silicon (the most common kind), one had to decide where it came from. If it came from the electronics industry, even if it is the off-grade rejected material, the energy cost is very high, as shown in Box 3.5. On the other hand, if the solar industry grows to the extent that it can build its own factories to produce solar-grade silicon of lower purity, the energy cost would be much lower. Both the worst-case and best-case scenarios for silicon were compared with CdSe and CIS (copper indium diselenide) thin-film systems. (CIS is similar to the CIGS men­tioned above.) The results for energy payback time are shown in Fig. 3.42.

Here is what went into these calculations. First, the materials used were listed. For thin film, these were glass, plastic, water, and the electronic layers. Glass was

by far the largest part, and the thin films the smallest. Then the energy to make these materials and the electricity to fabricate the cells in the factory were evaluated. That is for just the bare cell. To this must be added the balance-of-system; namely, the parts needed to complete functioning modules and arrays. These include aluminum for the frame, steel for the supports, cables and connectors, and the electric equip­ment for converting DC to AC at the grid voltage. There is also fuel oil used in installation. The energy cost of decommissioning and recovery of materials at the end of life was not included, but these were considered by Fthenakis [11]. The energy used was assumed to be the mix of fossil and hydro energy typical of Europe, with 32% average efficiency in generating electricity. As for the solar energy output, the assumptions were quite conservative. The sunlight available was 1,700 kWh/m2 per year, typical of southern Europe, not a desert. A 25% efficiency loss was assumed to account for dust accumulation and electrical equipment. The lifetime of the system was taken to be only 20 years.

The calculated energy payback times are shown in Fig. 3.42. As expected, it is very long for silicon in the worst case, when it is obtained from the electronics industry. However, if special factories are built to produce solar-grade silicon in ribbon form, the payback time is competitive with thin film. CdTe is the clear win­ner in this study, its payback time being only 1.5 years. The graph also shows the breakdown between the energy costs of the bare cells and the balance-of-system or BOS. Note that for CdTe cells, it is the BOS that takes the most energy to make. The global warming potential (CO2 emissions) of these systems is usually also calculated in these studies, and of course it is much smaller than that of fossil-fuel energy sources. After initial greenhouse gas (GHG) emissions during buildup, a solar plant produces electricity with almost no emissions for up to 30 years.

Cadmium is a very toxic element. In 2009, there was an uproar because some toys imported from China were found to contain cadmium in ingestible form. However, that does not mean that a compound like CdTe is toxic. Salt, NaCl, is certainly not dangerous although sodium and chlorine are themselves very toxic elements. In the case of CdTe, one worries that Cd could be emitted into the envi­ronment during manufacture and operation, even though the cells are encapsulated in glass and Cd is very stable, with almost zero evaporation. Unknown to most people, incidental emission of Cd also occurs in coal and oil plants. Raugei et al. [10] estimated the emission of Cd from a solar plant and found that it is 230 times smaller than from a coal plant for the same energy output! Detailed evaluations of dangers from toxic substances have been done by Fthenakis et al. [12, 13].

The amount of land used in solar power and the environmental impact on it has been compared with other energy sources by Fthenakis and Kim [14]. Not surpris­ingly, these solar proponents find that solar energy requires the least amount of land and biomass energy requires the most. The use of land in coal and nuclear power includes the land destroyed in mining and waste storage. Hydroelectricity uses dams which convert land into lakes. However, the usage of the area may actually be improved, and wildlife may only be changed from animals to fish. A large area covered with solar arrays may still allow desert animals, birds, and tortoises to live if some plants are allowed to grow under the panels. However, the reflectivity (albedo) of the desert will be decreased by the absorbing black panels. A very large area of these may affect cloud formation and the entire climate in the region.

The life-cycle studies of solar power are less complete than those of wind power and seem to be optimistic. The wind studies included replacement of parts as they wore out and the energy costs of inspection and maintenance, including the gasoline usage by the inspectors. Dust will cover the solar panels and should be washed off. In the desert, there is no water. The glass covers of the panels will be blasted in sandstorms. In temperate climates, plants will grow and have to be pulled out before they get too high. With no weeding, a solar farm will be immersed in a dense forest in 10-20 years. Space must be left open between rows of panels for machines to do this. The Mars rovers have experienced what happens to solar panels without main­tenance. Dust accumulated on them, decreasing their power. The Rovers depended on wind storms to blow the dust off. After years of dust accumulation, the power became so weak that communication became difficult. The rovers had to be manipu­lated onto a crater’s edge to tip the panels to face the sun more directly. In solar farms on earth, the panels are fixed.44 It has been estimated that mechanisms to track the sun would add 25% to the cost of the panels but could increase their capacity by 40%. The cost of energy storage for night time was not included in these studies.

However, the storage problem was addressed in admirable detail by Mason et al. [15]. The only method being considered is CAES, which is described in the Wind Energy section (Fig. 3.17). The electrical energy being stored is used to compress air into these caverns. When the energy is needed, this compressed air is released and used to help drive a gas turbine to produce electricity. CAES has been tested only at two places: in Germany, where a 290-MW plant has been operating since 1978, and in Alabama, where a 110-MW plant has been operating since 1991. These CAES systems were used to store excess electricity produced conventionally in off-peak hours. There are numerous sites in the USA where caverns suitable for CAES exist, but they cannot be close to the solar farms for several reasons. The deserts where there is the most sunlight have few suitable sites and insufficient water needed for cooling. They are also far from population centers. A system of high-voltage DC (HVDC) transmission lines is proposed to connect the solar plant to the storage plant. The energy capacities of the two plants also have to be matched.

The Mason study [15] considered a storage plant that provides peak power 10 hours a day, Monday through Friday when it is needed, and another for base­load power 24 hours a day for a future central-station solar farm. The daily solar output during the year was calculated, as well as the storage requirements during each day. The costs of the solar and storage plants were carefully itemized, includ­ing such items as maintenance, land preparation, interest during construction, and replacement of parts. The HVDC cost was included, as well as the substations for converting DC to AC. The results for a peak-load PV-CAES system are summa­rized in Fig. 3.43. The cost of electricity from PV systems with storage is compared with that from an advanced-cycle natural gas plant with carbon sequestration. In the next 10 years, the cost of the PV part is expected to go down, but the CAES part does not go down as much. It accounts for a third of the total cost. Solar electricity for peak loads, it appears, will be competitive with that from natural gas by 2020.

For base loads, however, PV-CAES electricity will cost $0.118/kWh, considerably more than the $0.076 and $0.087, respectively, from gas and coal plants, both with CCS. This is all conjectural, however, since the cost, safety, stability, and legal problems of underground storage have never been tested on a large scale.