7.1.1 Costs and Markets

One of the most encouraging aspects of the current PV scene is the steady reduction in costs. Continuing improvements in cell and module efficiencies are making a substantial contribution, but above all it is the sheer volume of production in state‐of‐the‐art factories using highly automated facilities that is driving down costs. Right back in Section 1.5, we introduced the “learning curve” concept to illustrate how, for a wide range of manufactured products, costs tend to fall consistently as cumulative production rises. Figure 1.14 showed that PV costs have fallen for more than two decades by around 20% for every doubling of cumulative production—and the trend continues. The long‐held, cherished ambition of the PV community to produce modules at “one US dollar per watt” was finally achieved in 2009 in the case of high‐volume thin‐film CdTe manufacturing, and by 2017, the module prices for most commercial technologies had fallen to $0.40–$0.70/W.

Of course the cost of a PV system also depends heavily on balance‐of‐system (BOS) components, and there are design, installation, and maintenance charges to consider. Fortunately, most of these are also falling broadly in line with cumulative PV production and today typically represent—as they have in the past—about half of total system costs.

The speed of market penetration by a new technology normally depends greatly on economics. Potential purchasers of grid‐connected PV systems, which have come to dominate the global market, wish to know how much generating solar electricity would cost them. For example, if you are considering installing a rooftop PV system, how does the cost of a unit of electricity (1 kWh) compare with the price charged by the local utility, and does it look like an attractive investment? In the case of stand‐alone PV systems, there are different criteria since grid electricity is not generally available as an alternative; comparisons are more likely to be made with diesel generators, and decisions affected by environmental concerns, including noise and pollution.

It is important to bear in mind that, in many cases, the installation of a PV system is not only about money. Companies may be concerned to demonstrate their green credentials, schools to educate and inspire their pupils, and individuals to “do their bit” to reduce carbon emissions. You may know someone who, instead of buying an expensive new vehicle, settled for a cheaper model that burns less fuel and spent the rest of the money on a rooftop PV system. For citizens in developed economies, it can be as much a lifestyle choice as a purely economic one.

As far as the economic case is concerned, Figure 7.2, although necessarily speculative, illustrates some important trends. Predicted costs of PV electricity are plotted up to year 2030 for electricity supplied by utilities to domestic customers in Europe (red curve) and for electricity generated by rooftop grid‐connected PV systems in various countries (orange, green, and blue curves). Most experts expect that the increasing global demand for energy, together with falling fossil fuel reserves, will result in real price rises for conventional electricity in the coming years. This is shown by the red curve, assuming an annual increase of 2.5% compound. By contrast, the price of solar electricity is expected to fall as cumulative PV production soars. In sun‐drenched European locations such as southern Spain and Italy (orange curve), the current cost is roughly competitive with conventional electricity because PV arrays are highly productive. In less sunny northern Germany and England (green curve), PV is expected to achieve “grid parity” by about 2020, and in Norway and Sweden (blue curve), perhaps 5 years later. But whatever the detailed timescales, the trends seem clear and inevitable—even if the citizens of northern Europe will need a bit more patience!

In many ways this picture is oversimplified. First, the costs of PV systems and the prices paid by consumers for grid electricity are not uniform between different countries. Second, price increases for grid electricity over the coming years cannot be predicted with any certainty. And additional factors will surely influence the cost of PV electricity—a cost that is by no means dictated solely by the choice of modules and the amount of sunlight. To understand this, we need to consider the capital and income components of a PV project; we will discuss the residential and commercial or utility cases separately.

Let us again imagine investing in a residential rooftop PV system. It is helpful to start by estimating expected cash flows over the life of the system, say, 20 years, as in Figure 7.3. This is the key ingredient of what is known as life‐cycle cost analysis.^{1, 2} Negative cash flows (expenditure) are shown red; positive ones (income) are shown blue. A major feature of PV systems is that the initial capital cost (A) produces by far the largest negative cash flow. This is followed by many years of positive cash flows representing the value of electricity generated (or savings due to electricity not purchased) and small negative ones to pay for routine system maintenance. Generally, it is also prudent to allow for additional capital expenditure to replace worn‐out or damaged BOS components such as charge regulators or inverters, or batteries in a stand‐alone system (B, C, and D). And finally we may hope to obtain an end‐of‐life scrap value for the system (E).

We are now in a position to assess the financial viability of the project. Of various measures, the easiest to understand are the simple payback period, the number of years it takes for the total costs to be paid for by the income derived from the system, and the rate of return, the percentage annual return on the initial investment. But it is hard to know how long the system will last or to allow for additional capital injections that may be needed as time goes by (items B, C, and D above).

An even more important limitation is that the simple payback period and rate of return take no account of the “time value” of money—a major consideration for a long‐term project. In a nutshell, a cash flow expected in the future should not be given the same monetary value today. For example, would you rather have $100 today, or the expectation of $150 in 10 years’ time? Your answer will probably depend on predicting future interest rates (you could put the money in the bank), or the confidence you have about future payments, or you may prefer to purchase something for $100 today. A proper life‐cycle cost analysis takes this into account by referring all future cash flows to their equivalent value in today’s money using a discount rate. This is the rate above general inflation at which money could be invested elsewhere, say, between 1 and 5%. In this way the present worth of a complete long‐term project can be estimated and compared with alternatives, allowing a more realistic investment decision to be made. As you may imagine, a positive value of present worth is generally taken as a good indication of financial viability.

So far so good, provided we recognize that the decision, even when based on careful life‐cycle cost analysis, contains uncertainties about technical performance, system and component lifetimes, interest rates, and the future price of electricity. And, as we have previously noted, it may also be based on environmental and social factors.

Investment in a commercial or industrial system carries less uncertainty as it is often linked with a power purchase agreement (PPA) with a utility that guarantees revenue that, over the course of the investment period, carries all initial and recurrent expenses. Solar as well as other renewable energy systems may have a high up‐front cost, but they use free solar or wind “fuel” during the many years they operate. On the other hand, the operating costs of coal and natural gas‐burning power plants are high and uncertain as they depend on the price and availability of fuel. Thus to compare the cost of energy technologies with different operating conditions, we need to sum the capital and operating costs and assign them to the amount of electricity produced over a period of time. For this purpose, the levelized cost of electricity (LCOE) is defined as a constant unit cost, per kilowatt‐hour or megawatt hour, of a payment stream over a period of time, which has the same present value as the total cost of constructing and operating a power‐generating plant over its life. It represents the constant level of revenue necessary each year to recover all expenses over the life of a power plant. In simple words, LCOE represents the total life‐cycle cost divided by the revenue from the total lifetime electricity production. The costs include the initial capital investment and the lifetime operating expenses, less the end‐of‐life value discounted into the present value. The revenue from the energy production depends on the location (solar irradiation), the performance ratio (PR), and the life of the system. A detailed discussion of these factors in the economics of PV systems is given by Campbell.³ A simplified LCOE equation is shown in the following text and a calculator can be found at the book resource website: www.wiley.com/go/fthenakis/electricityfromsunlight.

An important constituent in the LCOE is the cost of financing (debt service) of the relatively high equity investment needed up front. The schematic in Figure 7.5 illustrates this concept. The cost of financing is linked to the real and/or perceived risk of the asset. Long‐term PPA with utilities and government feed‐in‐tariff (FIT) programs reduce the risk to investors. An equally important consideration is that modern PV power plants are reliable, resilient, and utility friendly.4 Deployment of PV power plants removes the uncertainty of fuel supply and prices, and, as more field experience is gained, these plants are expected to last longer than thermoelectric power plants, because they lack the high‐temperature operating moving parts of steam and gas turbines. As PV grows and becomes more familiar, these attributes are more and more realized by the markets.

We have summarized the ideas behind conventional life‐cycle cost analysis, with its positive and negative cash flows and levelized cost of electricity that averages all costs over a certain period. But what if the economics are affected by a government decision to offer capital grants to offset the initial purchase price or suddenly to change or terminate grants that are presently available? And what if the price paid for renewable electricity is bolstered by special tariffs that may be altered or removed by a change of government? Over the years there have been many such stop–go incidents in countries as wide apart as Australia, Spain, and the United States. One of the biggest threats to rational decision making and steady growth in the PV market is uncertainty about government policy, and one of the biggest benefits is consistent long‐term support. We shall discuss support schemes in the next section.

You may be wondering why governments offer financial support to PV in the first place. There are two principal reasons. First, the products of a new high‐tech industry tend to be very expensive at the start, before cumulative production gathers pace. If governments wish to pursue urgent policy objectives such as the reduction of carbon emissions, they may decide to stimulate market development with financial incentives. Second, Figure 7.5 makes clear that PV, like other renewable energy technologies including wind and wave, has its major costs “up front,” with no fuel charges. This is quite different from conventional electricity generation based on fossil fuels. Projects with high initial costs that must be set against future income are commonplace for large corporations but tend to be far more problematic for small businesses, organizations, and individuals who find it hard to raise the initial capital.

Government support, although generally welcome and necessary for PV, tends to distort the market and prevents it from behaving according to the assumptions of classical economics. Realistic life‐cycle cost analysis becomes more problematic. In effect the global PV market becomes split into a number of sub‐markets with different characteristics. As an extreme example, the decision of an organization to install a large grid‐connected system on its office building is likely to be influenced by very different financial criteria and incentives from that of a family in a developing country struggling to find initial funds for a solar home system (SHS). This is not to say that economic analysis is worthless, just that it should be approached and interpreted with caution. If you refer back to some of the photographs in earlier chapters, you will see plenty of examples of PV systems based on a wide range of investment criteria—political, economic, environmental, and social.

7.1.2 Financial Incentives

We have already noted that PV, an exciting new technology with major environmental benefits, both justifies and deserves the support of governments wishing to accelerate market growth and counter the effects of air pollution and global warming. The solar PV market grew very rapidly in recent years, mostly driven by government policies supportive of renewable energy; these early policies created a significant market that catalyzed technological improvements that reduced costs. In turn, cost reductions enabled worldwide deployment and further cost reductions.⁵

Japan showed the way in 1994 with a 70 000 solar roofs program. Germany, after succeeding with its own 100 000 roofs program, went from strength to strength after 2004, thanks to improvements in its groundbreaking renewable energy legislation. Spanish government legislation led to an extraordinary burst of activity in 2008 when 2.7 GW_p of PV capacity was installed in a single year (you may like to refer back to Section 4.7 on large PV power plants). The United States, held back during the years of the Bush administration, surged ahead during the Obama years. In spite of a certain amount of stop–go in all these programs and difficulties due to the global economic recession, many other governments around the world have now joined the pioneers by offering substantial financial incentives to install PV systems.

Of the various ways in which governments have sought to provide financial incentives for the installation of grid‐connected PV systems, three key ones are particularly relevant to our discussion here:

• Capital grants, in the form of rebates and investment tax credits (ITCs), to offset the initial cost of the system
• Special tariffs for the electricity generated, which is either used on‐site or fed into the grid
• Financing options and loan warrantees

7.1.2.2 Special Tariffs

The second approach, tariffs for the electricity generated, increases the amount of income received over the life of the system. It therefore encourages the purchase of high‐quality systems that are carefully installed and maintained. Often taking the form of Feed‐in Tariffs (FITs), the subsidies are financed by requiring utilities to buy renewable electricity at above normal market prices. The tariff is based on the cost of the electricity plus some reasonable return for the investor. From a government viewpoint, FITs are generally “revenue neutral.” The cost is spread over all customers who must pay a small annual percentage increase in their electricity bills. Their major advantage is the guaranteed income payments offered over timescales of 20 or 25 years, reducing uncertainty and increasing investor confidence.

FITs are implemented in many jurisdictions including EU countries, Australia, Brazil, Canada, and China and in California. The FIT has played a major role in boosting solar energy in Germany and Italy. In Germany, a renewable energy law passed in 2000 introduced a FIT that proved extremely effective at stimulating a range of renewable energies. The PV tariffs were tweaked in 2004 to compensate for the termination of the German 100 000 roofs program, providing payback times of around 8–10 years. This resulted in a veritable boom in PV installations. Huge numbers of PV arrays were put on domestic and commercial buildings, farmers placed PV on barns and in fields, and many large PV power plants were commissioned. In 2005 total installed capacity in Germany exceeded 1 GW_p, and in 2008 it reached 6 GW_p. Of course, a generous FIT can become unsustainable if continued too long, so in many cases tariffs for new installations are lowered by a certain percentage each year to take account of PV’s expected “learning curve.” Guaranteed payments over a period of years offer investors sufficient confidence to fund the initial capital in solar projects. However, the challenge with FIT designs is that it can over‐ or under‐compensate solar projects as there is no market pricing mechanism in many programs. As the costs of FITs usually are passed on to consumers, it is essential to design incentives that attract sufficient investment while yet permitting adjustment of subsidies for new additions to capacity as technology costs fall, so avoiding unnecessary increases in electricity prices and maintaining public acceptance.

An extra generous Spanish FIT started in 2004 propelled Spain on an exciting journey into the gigawatt era. However, a few years later, the Spanish government introduced annual caps, and this dampened a market that had surged beyond expectation.

More than 60 other countries have now entered the FIT arena, and many are no doubt learning from the operational experiences of the pioneers. And in spite of the negative effects of the global economic recession that started in 2008, most commentators believe that PV and other renewable energy technologies will ride the storm relatively unscathed and continue to attract the support of governments increasingly focused on the dangers of global warming.

7.1.3 Rural Electrification

So far we have been concentrating on the economic aspects of grid‐connected systems and the ways in which governments in developed nations encourage the development of PV markets. Passing reference has been made to stand‐alone PV systems, noting that the chief competitor for supplying electricity in remote areas is generally the diesel generator. But all this relates to relatively wealthy nations including those that have driven PV’s spectacular growth over the last decade.

There is another important dimension to the terrestrial PV story, and it concerns the provision of relatively small amounts of solar electricity to families and communities in the developing world, who have little prospect of buying and maintaining diesel generators and no prospect of connection to a conventional electricity grid in the foreseeable future. This challenging yet worthwhile activity is referred to as rural electrification.

A major aspect of rural electrification is the supply of SHSs to individual families, and we shall concentrate on it here. Other applications include irrigation and water pumping (see Section 5.5.3), refrigeration of vaccines and medicines in remote hospitals, and the supply of PV systems to small businesses and institutions. It is sobering for those of us who live in developed countries to realize just how little electricity is needed to provide valuable services to people who otherwise have none. For example, the average electricity consumption of a household in Western Europe is around 10 kWh/day. The stand‐alone system for a holiday home that we designed in Section 5.4.1 (see also Figure 5.15) assumed a consumption of 2.2 kWh/day, sufficient to run a good range of modern electrical appliances if used with care. But when we consider an SHS based on a single PV module, typically rated between 30 and 60 W_p, the figure is more likely to be 0.2 kWh/day—one‐fiftieth of the electricity taken for granted by most European families. This modest amount can power a few low‐energy lights and a small TV, offering genuine improvements to rural living standards and a contact with the wider world (Figures 7.9 and 7.10).

Like other PV systems, SHSs have most of their costs up front. A system comprising a small PV module, charge controller, 12 V battery, cabling, switches, and some low‐energy lights may retail for a few hundred US dollars or euros. This may not seem much in America, Australia, or Europe, but to many families engaged in subsistence farming in the developing world, it looks like an unattainable fortune. The SHS market—or perhaps we should say “markets,” because conditions vary widely from one country to another—therefore needs its own financing arrangements.

Efficient and convenient lighting is arguably the most important service offered by an SHS. Families without electricity often spend a substantial proportion of their disposable monthly income on kerosene lamps, candles, or dry cell batteries, so this money is in principle available to pay for a PV system (Figure 7.10).

Typical financing schemes for SHSs include:

• A short‐term loan to cover all or most of the initial cost, paid back with interest over a period between 1 and 3 years
• A leasing arrangement whereby an SHS is installed and maintained by an organization or company in exchange for monthly fee‐for‐service payments

A wide variety of official, commercial, and aid organizations are involved in the financing of SHS programs around the world. In addition to national governments, local banks, and leasing companies, the United Nations and the World Bank are actively involved and so are various nongovernment organizations (NGOs) and aid agencies.

Among the many countries with impressive rural electrification and SHS programs we might mention the following:

• In Asia: China, India, Sri Lanka, Bangladesh, Thailand, and Nepal
• In the Americas: Mexico, Brazil, Argentina, Bolivia, and Peru
• In Africa: Morocco, South Africa, Kenya, and Uganda

We end this section with a few comments on cultural and social issues surrounding the introduction of high‐tech products into developing countries. In many cases a small stand‐alone PV system represents the only contact of a rural family with 21st‐century technology (Figure 7.11). Proper system maintenance can be a problem, and education is a very important part of the package. Although PV modules are normally very reliable, the lead‐acid batteries used in SHSs need regular topping‐up and occasional replacement, modules must be kept reasonably free of dust and bird droppings, and electrical connections must remain tight and corrosion‐free. Such tasks are far removed from the experience of many rural communities (Figure 7.12). When SHSs are financed as part of a community electrification project, there may be problems of management and control. A great deal has been learned over the past 30 years about the cultural pitfalls of rural electrification, where failures tend to occur for reasons other than the purely technical. But any such difficulties should not detract from the great social benefits of rural electrification, which is surely one of PV’s most admirable achievements. The challenge remains for providing affordable and reliable electricity to 1.5 billion people in developing countries (Figure 7.13).

7.1.4 External Costs and Benefits

PV technologies carry one very important existential attribute that often is neglected in discussions of energy policy; namely, by displacing conventional fossil fuel‐based power generation, they prevent a spectrum of acute and chronic health and environmental impacts that carry a great economic cost to society as a whole. Monetization of the external costs of energy life cycles, including those of PV, is well documented in reports by the National Research Council (NRC), the Harvard School of Public Health, and others. These studies show that the greatest health and environmental effects in the life cycle of coal power generation are those from toxic air pollutants during combustion, followed by the impacts of mining, and GHG emissions. More specifically, the Harvard study estimated that the total external costs of coal‐fired electricity during extraction, transport, processing, and combustion are $345 billion or 18 cents/kWh of electricity produced. These estimates are based on health costs, health insurance, and damage costs that are not included in the electricity costs but are paid by the society at large. If these were fully accounted for, then the price of electricity from coal in the United States would have been much higher than the LCOE of PV. Carbon dioxide capture with carbon sequestration (CCS) is advocated widely, enabling the continuation of coal burning for power generation, but CCS, while reducing GHG emissions, will lead to increases of toxic emissions and environmental health and safety (EHS) impacts in mining regions, as coal consumption per unit of electricity output would increase. The same applies to the proliferation of natural gas from gas‐shale resources, as hydraulic fracturing increases the impacts of extracting gas on both the air and water pathways, and conversion to LNG for exportation further increases the upstream safety risks.

7.1.5 Policy Recommendations for Further Growing Solar Energy

The growth of solar PV industry will be driven by a multi‐year sustained policy mix that guarantees attractive returns on solar PV investments and addresses the technical and regulatory requirements for solar energy. This mix comprises FIT, federal ITCs, loan guarantees, RPS, and REC. This mix can be deemed successful by looking at the recent increase in solar PV deployment in developed countries. However, when trends in the solar PV industry are analyzed further, it becomes evident that there are issues with market stability and some gaps in the existing policy mix, hindering the further deployment of solar PV. The needed improvements in the federal policy mix include R&D funding, solar financing flexibility, reduction of permitting burdens/costs, markets for clean energy credits, and imposing fees for carbon emissions.³

7.2.1 Raw Materials

We start with the issue of raw materials. Most of the PV module is glass and most of the mounting structure is galvanized steel and aluminum. One point is clear: insofar as PV’s future is majorly based on silicon solar cells encapsulated in glass sheets, there is no problem. Silicon is one of the most common elements in the Earth’s crust and, almost literally, as plentiful as sand on the beach. There is no future scenario in which it could become exhausted, and the only concern is that it requires a lot of energy to process it to the high purity and crystallinity required for solar cells. This energy was initially supplied by hydroelectric power, but there is not enough hydro to support the growth we are currently experiencing, so unfortunately dirty fossil fuel energy is increasingly used in silicon purification. However, the energy return (from PV electricity when the system operates) is 10–30 times higher than the energy invested in the manufacturing of Si PV, so the processing is energetically sustainable. A residual concern is logistical and financial; although the raw materials are abundant, advance scheduling is needed for both silicon and glass manufacturing plants as it takes at least three years for such large plants to be completed. Our studies indicate that at a 30% annual growth, the glass needs of the PV industry by 2025 will be higher than the current total flat glass manufacturing capacity (about 8 billion square meters in 2015).⁶

However, there have been concerns that physical constraints in the availability of some materials may limit the growth of thin‐film PVs, which lead the industry in cost reductions. In 2012 European Commission and US DOE reports listed gallium, indium, and tellurium (Ga, In, Te), which are used in CIGS and CdTe PV, as critical in terms of supply risk and economic importance. Their use is expected to increase because the entire PV industry is experiencing high growth. Furthermore, beyond PVs, the usage of gallium in integrated circuits and optoelectronics and indium in flat‐panel displays is expected to rise. These materials are limited in supply because they are minor by‐products of aluminum, zinc, copper, and lead production; accordingly, their production is inherently linked to that of the base metals, and thus, the rate of production of these base metals must be examined. The energy to extract the elements may pose an additional limitation.

Copper is the parent metal for tellurium; zinc for indium, germanium, and gallium; lead for tellurium, cadmium, and indium; and aluminum for gallium. The demand for Cu is expected to reach a peak within 50 years, and for Zn and Pb in about approximately 20 years, whereas the demand for Al is forecasted to increase through the end of the century. The US Geological Survey predicted a rate of growth in global demand for copper of 3% per year between 2000 and 2020; so far, this prediction has been correct. Most models predict a peak in copper production in 2050–2055; thereafter, demand is expected to decrease gradually or remain about constant during the rest of the 21st century, as the role of recycling becomes more significant. Zinc extraction grew by 3.2% annually between 1910 and 2002, and this trend has continued over the last 10 years. The demand growth rate for Zn till 2030 is assumed to be the same as for Cu.

Let us now discuss the production of the so‐called critical metals (Te, In, and Ga), which are by‐products of the base metals.¹ The main sources of tellurium are the anode slimes from copper electrorefining operations, and their average recovery rate is about 40%. In contrast, the recovery rate of copper from the same ores is 80% or better, and that of gold is over 95%. Evidently, the market drives the rate of recovery, with a higher demand and price justifying additional processing. Nothing inherently prevents recovery rates for tellurium from being as high as those for copper, or perhaps even gold, provided that the price is sufficiently high. Indeed, the concentration of gold in anode slimes typically is lower than that of tellurium.

However, there is a limit to the price of tellurium that will sustain affordable CdTe PVs. At US$120/kg, the tellurium currently used in CdTe modules is approximately US$0.01/W_p. Several scenarios are suitable for assessing the future availability of tellurium. All are related to projected copper production because, with very few exceptions, the quantities and prices of the minor metals do not warrant the extraction and processing of ores without the simultaneous recovery of copper. Starting with the tellurium content in copper anodes of 1250 metric tons (tonnes) per year and applying 3.1% annual growth and a gradual increase to 80% recovery from anode slimes, by 2020, the annual primary production of metallurgical‐grade tellurium would be 1450 tonnes/year. Figure 7.15 shows the mass of Te available for PV, assuming that the current demand for other than PV applications remains constant. In addition to tellurium from copper mines, there are other types of smaller reserves, including tellurium‐rich mineral deposits in China and Mexico from which the near‐term direct mining of tellurium is economically feasible. Over the longer term, tellurium recovery from mining tailings and from refining of lead–zinc ores is also possible. In addition, massive resources of tellurium exist in ocean‐floor ferromanganese nodules, reportedly as much as 9 million tonnes at mean concentrations of 50 ppm. However, because quantitative information is not available for the former and because the recovery of metals from deep ocean is not currently cost‐efficient, these resources are not included in current assessments.

A secondary yet rich source of Te is found in end‐of‐life CdTe modules that contain up to 500 ppm of Te. The industry is already practicing Te recovery from defect and field returns of CdTe modules. The technical and economic feasibility of recycling most materials from CdTe PV is proven. Laboratory‐ and pilot‐scale studies by Fthenakis and Wang at Brookhaven National Laboratory (BNL) achieved 99.99% separation of tellurium and cadmium from end‐of‐life modules at an estimated cost of US$0.02/W_p. On an industrial scale, a 90% overall recovery rate is reported from First Solar. The collection of spent modules can be expected to be 100% from large utility installations and 80% from residential installations, provided that there are economic incentives or laws regulating disposal. Accordingly, recycling can become a significant source of secondary tellurium before the middle of the century (Figure 7.15).

CdTe PV production: The total annual production of CdTe PVs that tellurium availability in copper smelting can support will be constrained to 16–24 GW_p by 2020, 44–106 GW_p by 2050, and 60–161 GW_p by 2075 (Figure 7.16(a)). The tellurium‐based limit of cumulative global production of CdTe PVs (Figure 7.16(b)) is 120 GW_p by 2020, rising to 1–2 TW_p by 2050 and 4–10 TW_p by 2100. These limits are based strictly on tellurium coproduction during copper production from known resources and do not include the potential for direct mining or the discovery of additional resources.

For the production of CdTe PVs to continue growing by 40% per year, tellurium recovery from anode slimes must increase to 80%, for which there is already a technological basis. We also assumed increase in the efficiency of the modules to 16% and decrease of the thickness of the CdTe film thickness to 2 µm.

Indium: Indium is a by‐product of zinc extraction. Interestingly, it is not an especially rare element in the Earth’s crust; it is actually about three times more plentiful than silver but only extracted at 1/60th the rate, emphasizing the dependence of indium volumes on the scale of zinc mining. The supply of indium is tied to the production of zinc and is likely to remain so in the future. The price of indium reached a high of US$1000/kg in 2005 but is currently about US$600/kg. In 2010, the estimated production of indium was 1345 tons, of which 480 came from mining and another 865 tons came from recycling of used indium sputtering targets; by 2015 the annual production has increased to 1612 tons. The main use of indium is in liquid‐crystal displays (LCDs), accounting for 65% of current consumption; PVs use less than estimated 5% of the primary production of indium. Competing applications of indium present extra challenges to PV growth, and further, its recovery from the zinc circuit is already high (~70–80%), leaving little room for enhancement.

Gallium: Most gallium is produced as a by‐product of treating bauxite ore to extract aluminum; about 10% originates in sphalerite (ZnS), and it is produced during the purification stages of the zinc production circuit. The world resources of gallium in bauxite ore are estimated to be about 1 Mt, but evaluations of the reserves (deposits that currently can be mined economically) are lacking. Most gallium is extracted electrolytically from a solution of crude aluminum hydroxide in the Bayer process for producing alumina and aluminum. In 2010, the production of gallium was estimated to be 207 tons, of which 100 tons was derived from mining and the rest from recycling scrap. Only approximately 10% of alumina producers extracted gallium, with the others not finding it economical. The price of gallium reached a peak of US$2500/kg in early 2015 and is currently US$500–600/kg. Presently, almost all of the gallium produced is used in integrated circuits and optoelectronics, with both usages exhibiting upward trends. The estimated 2015 supply is 325 tons.

CIGS production: Under the listed assumptions on indium and gallium availability, the material‐constrained growth potential of CIGS PVs has been calculated as 13–22 GW_p/year by 2020, 17–106 GW_p/year by 2050, and 17–152 GW_p/year by 2075. These estimates assume 80% extraction recoveries and use of only 50% of the growth in the supply of indium for CIGS PVs, as well as foreseeable improvements in module efficiency and material requirements. Note that the estimates for midcentury and beyond are based on the presumption that the growth of zinc extraction will follow that of copper; this is questionable because the depletion time of zinc may be shorter than that of copper. Furthermore, recovering indium and gallium from CIGS is more challenging than recouping tellurium from CdTe, as their respective concentrations are lower and their separations are more difficult.

In summary, the availability of Te, In, and Ga indeed constrains the production of CdTe and CIGS PVs. However, several comprehensive assessments have shown that there are sufficient resources to bring each of these technologies to terawatt production by midcentury; if recycling is widely adopted, such production can reach up to 10 TW by the end of the century. Recycling the end‐of‐life PVs is becoming increasingly important as PV deployment grows. It helps in keeping the material costs low as it provides significant secondary resources at a price lower than the primary ones, it displaces energy from material production, and it resolves concerns about potential environmental contamination from the uncontrollable disposal of PV.

It is probable that other types of cells currently in the research phase, or entirely new ones not yet discovered, will be in volume production by 2050; so any aspirational projections of PV growth, for example, the Grand Solar Plan discussed in the first chapter, could be met with a mix of PV technologies including thin films, silicon, and probably new ones. The important issue is to prepare the infrastructure for hosting these technologies, for example, to invest in grid upgrades and long‐distance transmission instead of gas pipelines!

7.2.2 Land Use

PVs offer advantages for distributed power generation, and rooftop installations, which do not use any extra land, representing one‐half of today’s world market. However, ground‐mount PV power plants occupy significant amounts of land. Data from many PVPS in the United States show land use in the range of 5–8 acres/MW_dc (Figure 7.17). This is the land that modules occupy to receive the solar irradiation, thus the free “solar fuel,” plus the open areas needed to prevent shading, and the access roads and facilities. Now is this a lot of land? To answer this question we would need to make comparisons with the land used by the coal, natural gas, and nuclear life cycles.

Fossil fuel‐based generation such as coal has a seemingly low land footprint, as power plants take up a relatively small surface area for its large power output. However, the picture changes when life‐cycle land use is assessed, accounting for the direct (mining and fuel processing, plant footprint) and indirect (land usage for materials and building infrastructure needed to operate the mines) land transformed. The life‐cycle land usage for different sources of electricity can be compared on a “surface area per energy unit” basis, although in that case it is important to define a finite time scope as the land used for solar and wind could virtually generate electricity forever (in the case of this study: 30 years, the PV system lifetime). Land usage for PV and wind would perhaps better be described with a “surface area per power unit” as their power source for that surface area will exist for virtually infinite time. In contrast, the energy source from surface coal mines is exhausted after the fuel is extracted.

Historical data show that ground‐mount solar farms in the United States often use less land during their life cycle than coal during its life cycle (Figure 7.18).7

For the coal fuel cycle, the direct land transformation is primarily related to the coal extraction, electricity generation, and waste disposal stages, while the indirect land use refers to the upstream land use associated with energy and materials inputs during the fuel cycle. Land‐use statistics during coal mining vary with factors including heating value, seam thickness, and mining methods. Surface mining in the Western United States tends to disturb less area (per unit coal mined) than in other areas due to the thick seams, 2–9 m. Central states where seam thickness is only 0.5–0.7 m transform the largest area for the same amount of coal mined (Figure 7.19). On the other hand, underground mining transforms land for the most part indirectly. Wood usage for supporting underground coal mines accounts for the majority of indirect land transformation. Currently in the United States, about 70% of coal is mined from surface.

For operating a coal power plant, land is required for facilities including powerhouse, switchyard, stacks, precipitators, walkways, coal storage, and cooling towers. The size of a coal power plant varies greatly; a typical 1000 MW capacity plant requires between 330 and 1000 acres, which translates into 6–18 m²/GWh of transformed land based on a capacity factor of 85%. Another study based on a 500 MW power plant located in the Eastern United States estimated 32 m²/GWh of land transformation. Also, coal‐fired power plant generates a significant amount of ash and sludge during operation. Disposing the solid wastes in the US account for 2–11 m²/GWh, 50% each for ash and sludge.

The natural gas life cycle would use less land, but when it is extracted from the dilute formations on shale rocks, then the land occupation is expected to be as large as that of coal.

In addition to the use of the land, we should consider its transformation and possible damage. PV installations do not damage the land (Figure 7.17). Once the installation is completed, PV operation will not disturb the land, in contrast to coal mining, which often pollutes the land. Also fossil or nuclear fuel cycles need to transform certain amount of land continuously in proportion to the amount of fuel extracted. Restoring land to the original form and productivity takes a long time and often is infeasible. Accounting for secondary effects including water contamination, change of forest ecosystem, and accident‐related land contamination would make the PV cycle even better than other fuel cycles. For example, water contamination from coal and uranium mining and from piles of uranium mill tailings would disturb adjacent lands. Additionally, land transformed by accidental conditions especially for the nuclear fuel cycle could change the figures dramatically. The Chernobyl accident contaminated 80 million acres of land with radioactive materials, irreversibly disturbing 1.1 million acres of farmland and forest in Belarus alone.

We now turn to the question of how much land is needed for a transition to a solar‐based power infrastructure. This has already been mentioned in Section 1.5, where we suggested that an area of land 140 × 140 km, or 20 000 km², roughly 3 times the size of London or Paris, would be sufficient to accommodate 1000 GW_p of PV modules. It seems that by 2020, or soon after, we may be approaching this huge total, some 3 times greater than global installed capacity of 350 GW in 2017, assuming that PV continues its present remarkable progress. But where would the land actually come from, and would we resent it?

If 20 000 km² sounds like a large parcel of land, consider some even larger ones: the Sahara Desert is about 850 times bigger, the Australian Outback about 200 times, and the state of Arizona about 15 times. In the United States, cities and towns cover some 700 000 km², and in many countries wide tracts of land are set aside for military uses, airports, highways, fuel pipelines, and so on. In short, if the world’s PV is sensibly spread around among the world’s nations, the landscapes seen by the vast majority of people will be virtually unchanged from those they enjoy today.

Of course this is far from the whole story, because PV can be installed on buildings (Figure 7.20). There are vast numbers of existing homes, offices, public buildings, factories, warehouses, airports, parking lots, and railway stations with suitable roofs and façades, and we may be sure the that tomorrow’s architects will be even more aware of the possibilities. BIPV will undoubtedly provide a major part of PV’s future space requirements, leaving deserts and other unproductive land to supply most of the balance. Sunshine is everywhere, high and low, city and country, and at fairly predictable levels.

7.2.3 Water Use

PVs have a major advantage against any thermoelectric power generation (e.g., coal, natural gas, nuclear, biomass, concentrated solar power) as it does not need any water for power generation. Electricity generation via conventional pathways accounts for a major part of water demand. According to US Geological Survey, thermoelectric power plants in the country accounted in 2009 for 49% of the freshwater withdrawal, much higher than withdrawals for agricultural irrigation that accounted for 31% of the total water demand. Most of this water is used for cooling during plant operation, and a large fraction is returned, at higher temperature, into the sources from where it was extracted. In contrast, renewable energy sources, such as PV and wind power, do not use water during their operation with the exception of water for cleaning dust from high‐concentration PVs, an amount negligible in comparison with the water demand in thermoelectric power plant operation (Figure 7.21). However, every energy‐generation technology does use water sometime throughout its entire life cycle. For example, during the PV life cycle, water is used for cleaning silicon wafers and glass substrates and preparing chemical solutions. In addition, a significant amount of the electricity used to purify silicon and other semiconductor materials is generated by thermoelectric power plants that rely on a water‐cooling system. Conversely, as well as using water during their operation, such plants need water both directly and indirectly during fuel acquisition, plant construction, and disposal stages. The total water withdrawal during the life cycle of PV was estimated to be about 350 m³/GWh for PV operation in SW United States, compared with 2 500–120 000 m³/GWh for various coal, nuclear, and national gas plants in the region. Overall, it was estimated that PV deployment in SW United States would displace 1700–5600 m³/GWh of water demand, when it displaces grid electricity.⁸

The water issue is region specific and is becoming increasingly important as climate change materializes with increased drought episodes.

7.3.1 Life‐Cycle Analysis

In the previous section we considered PV’s requirements for raw materials and land—two environmental issues that surface before PV production even begins. Further important environmental questions arise during a PV system’s lifetime, which starts with extraction and purification of raw materials; proceeds through manufacture, installation, and many years of operation; and ends with recycling or disposal of waste products. The whole sequence is referred to as a life cycle, and it is important to appreciate its environmental consequences. Note that this form of life‐cycle analysis (LCA) is not the same as the classical economic life‐cycle costing version previously introduced, which deals with cash flows and financial decisions. We are now moving on to something much broader, with important implications for global energy policy and society as a whole.

In this brief introduction we will consider LCA under two main headings:

• Environmental and societal costs. What costs, in addition to classic economic costs, are incurred or avoided?
• Energy balance. How does the amount of electrical energy generated over a system’s lifetime compare with the energy expended in making, installing, and using it?

We start with environmental and societal costs. It is clear that all methods of energy production—whether based on oil, gas, coal, nuclear, or renewable sources—have impacts on the environment and society at large that are ignored by the traditional notion of “cost.” A narrow economic view of industrial processes assesses everything in terms of money while ignoring other factors that common sense tells us should be taken into account in any sensible appraisal of value. For example, the “cost” of generating electricity in nuclear power plants has traditionally been computed without taking any account of accident or health risks; and in the case of coal‐fired plants, without acknowledging their unwelcome contribution to global warming; and in the case of wind power, without placing any value on landscape.

There are two main reasons for this apparent shortsightedness. First, aspects such as health, safety, environmental protection and the beauty of a landscape cannot easily be quantified and assessed within a traditional accounting framework. We all know they are precious, and in many cases at least as important to us as money, but appropriate tools and methodologies for including them are only now being developed and accepted. It is surely vital to do this, because so many of our current problems are bound up with the tendency of conventional accounting “to know the price of everything and the value of nothing.”

The second reason relates to the important notion of the external costs of energy generation. These costs, most of which are environmental or societal in nature, have generally been treated as outside the energy economy and to be borne by society as a whole, either in monetary terms by taxation or in environmental terms by a reduction in the quality of life. They contrast with the internal costs of running a business—for buildings and machinery, fuel, staff wages, and so on—that are paid directly by a company and affect its profits. If planet Earth is treated as an infinite “source” of raw materials and an infinite “sink” for all pollution and waste products, it is rather easy to ignore external costs. For example, it seems doubtful whether the 19th‐century pioneers of steam locomotion ever worried much about burning huge quantities of coal or the 20th‐century designers of supersonic civil airliners about fuel efficiency and supersonic bangs. One of the remarkable changes currently taking place is a growing world view that external costs should be worked into the equation—not just the local or national equation, but increasingly the global one. In other words, external costs should be internalized and laid at the door of the responsible industry or company. In modern phraseology, “the polluter should pay.”

Many of the external and internal costs associated with industrial production are illustrated in Figure 7.22. The external ones, representing charges or burdens on society as a whole, are split into environmental and societal categories, although there is quite a lot of overlap between them. You can probably think of some extra ones. Internal costs, borne directly by the organization or company itself, cover a very wide range of goods and services, from buildings to staff wages. The distinction between internal and external costs is somewhat clouded by the fact that many items bought in by a company, for example, fuel and materials, have themselves involved substantial “external” costs during production and transport. In the case of electricity generation, a proper analysis of the environmental burdens should take proper account of all contributing processes and services “from cradle to grave,” whether conducted on‐ or off‐site. Needless to say, this is a challenging task.

One of the special difficulties facing renewable electricity generation, including PV, is that so many of its advantages stem from the avoidance of external costs and are therefore hidden by conventional accounting methods. Renewables tend to produce very low carbon dioxide emissions, cause little pollution, make little noise, create few hazards to life or property, and have wide public support. PV can claim all these benefits. Yet when economists and politicians talk about PV, reduction or avoidance of external costs is seldom mentioned. Fortunately, energy experts and advisers to governments are taking increasing notice of environmental LCA in their decisions and assessing the risks and benefits of competing technologies on a more even footing. Certainly, the PV community must be involved in countering outdated thinking about the wider benefits of its technology.

We now move on to the much‐discussed topic of energy balance. Clearly, it takes energy to produce energy. But how does the total amount of electrical energy generated by a PV module or system over its lifetime actually compare with the input energy used to manufacture, install, and use it? Closely related to the energy balance is the energy payback time (EPBT), the number of years it takes for the input energy to be paid back by the system. We naturally expect PV to have favorable energy balances and payback times, especially in view of its claims to be clean and green.

Two initial points are worth making. Firstly, energy payback is not the same as economic payback. The latter is concerned with repaying a system’s capital and maintenance costs (including cost of energy consumed) by a long‐term flow of income and is essentially a financial matter; energy payback is much more about the environment. Secondly, the environmental benefits of a short payback time depend on the present energy mix of the country, or countries, concerned. If the required input energy is largely derived from coal‐burning power plants, it is more damaging than if it comes from, say, hydroelectricity.

Major energy inputs to a PV system occur during the following activities:

• Extraction, refining, and purification of materials
• Manufacture of cells, modules, and BOS components
• Transport and installation

Interestingly, some of the most significant energy inputs are for components such as aluminum frames and glass for modules and concrete foundations for support structures in large PV plants. Although the energy required to refine pure silicon and make crystalline silicon solar cells is considerable, the continual trend toward thinner wafers using less semiconductor material is reducing this problem. The energy input for thin‐film cells is generally very small. Also, in most ground‐mount installations, concrete foundations are not used any longer as supports are pinned into the ground.

The other side of the energy balance—the total electrical energy generated by a system over its lifetime—depends on a number of factors discussed in previous chapters:

• Efficiency of PV modules and other system components
• The amount of annual insolation
• Alignment of the PV array and shading (if any)
• The life of the system

Now, let us discuss the LCA in a more structured way.

The energy balance is most favorable for systems that are efficiently produced in state‐of‐the‐art factories and installed at optimal sites in sunshine countries. Things get even better if systems last longer than their projected or guaranteed lifetimes—but of course this is hard to predict. Some life‐cycle studies carried out in the early years of the new millennium painted a rather gloomy picture of PV’s environmental and health impacts, due largely to the fossil fuel energy used during cell and module manufacture. However, up‐to‐date peer‐reviewed studies that take proper account of advances in PV engineering have corrected this picture. The tremendous advances in improving the energy consumption and production of modern PV systems are illustrated at the steep decline of EPBT shown in Figure 7.23. EPBTs of current PV power plants are between 0.5 and 2 years, depending on the radiation in the place of installation and the type of technology used; this is a two‐order‐of‐magnitude improvement from the EPBT of the early (1970) systems, and the picture is getting even better.⁹

To better understand this result, we need to discuss the elements of the LCA, which is a comprehensive framework for quantifying the environmental impacts caused by material and energy flows in each and all the stages of the “life cycle” of a product or an activity. It describes all the life stages, from “cradle to grave,” thus from raw material extraction to end of life. The cycle typically starts from the mining of materials from the ground and continues with the processing and purification of the materials to manufacturing of the compounds and chemicals used in processing and manufacturing of the product, transport, installation if applicable, use, maintenance, and eventual decommissioning, and disposal and/or recycling. To the extent that materials are reused or recycled at the end of their first life into new products, then the framework is extended from “cradle to cradle.” This life cycle for PV is shown in Figure 7.24.

The life‐cycle cumulative energy demand (CED) of a PV system is the total of the (renewable and nonrenewable) primary energy harvested from the geo‐biosphere in order to supply the direct energy (e.g., fuels, electricity) and material (e.g., Si, metals, glass) inputs used in all its life‐cycle stages (excluding the solar energy directly harvested by the system during its operation). Thus,

(7.1)

where

The CED of a PV system may be regarded as the energy investment that is required in order to be able to obtain an energy return in the form of PV electricity.

The life‐cycle nonrenewable cumulative energy demand (NR‐CED) is a similar metric in which only the nonrenewable primary energy harvested is accounted for; details are given in the IEA PVPS Task 12 LCA Guidelines report.¹⁰

EPBT is defined as the period required for a renewable energy system to generate the same amount of energy (in terms of equivalent primary energy) that was used to produce (and manage at end of life) the system itself:

(7.2)

where E_mat, E_manuf, E_trans, E_inst, and E_EOL are defined as previously mentioned; additionally,

η _G[MJ_el/MJ_PE‐eq]: grid efficiency, that is, the average life‐cycle primary energy to electricity conversion efficiency at the demand side

For systems in operation for which records exist, the annual electricity generation (E_agen) is taken from the actual records. Otherwise it would be estimated with the following simple equation (note the units into parentheses):

where

• Irradiation is the global irradiation on the plane of the PV.
• Module efficiency is the manufacturer rated efficiency measured under 1 kW_p/m² irradiance.
• The PR (also called derate factor) describes the difference between the modules’ (DC) rated performance (the product of irradiation and module efficiency) and the actual (AC) electricity generation. It mainly depends upon the kind of installation. Mean annual PR data collected from many residential and utility systems show PR or 0.75 and 0.80, respectively.

In general, the PR increases with (i) decline in temperature and (ii) monitoring the PV systems to detect and rectify defects early. Shading, if any, would have an adverse effect on PR. This means that well‐designed, well‐ventilated, and large‐scale systems have a higher PR.

Using either site‐specific PR values or a default value of 0.75 is recommended for rooftop and 0.80 for ground‐mounted utility installations; these default values include degradation caused by age.

When site‐specific PR values based on early years’ performance are used, degradation‐related losses should be added to longer‐term projections of the performance.

E _agen is then converted into its equivalent primary energy, based on the efficiency of electricity conversion at the demand side, using the grid mix where the PV plant is being installed. Note that E_agen is measured (and calculated) in units of kWh and we first have to convert it to MJ (1 kWh = 3.6 MJ) and then use η_G to convert megajoules of electricity to megajoules of primary energy (MJ_PE‐eq). Thus, calculating the primary energy equivalent of the annual electricity generation (E_agen/η_G) requires knowing the life‐cycle energy conversion efficiency (η_G) of the country‐specific energy mixture used to generate electricity and produce materials. The average η_G for the United States and Europe are, respectively, approximately 0.30 and 0.31.

Energy return on (energy) investment (EROI) is defined as the dimensionless ratio of the energy generated over the course of its operating life over the energy it consumed (i.e., the CED of the system). The electricity generated by PV needs to be converted to primary energy so that it can be directly compared with CED. Thus EROI is calculated as

where T is the period of the system operation; both T and EPBT are expressed in years.

EROI and EPBT provide complementary information. EROI looks at the overall energy performance of the PV system over its entire lifetime, whereas EPBT measures the point in time (t) after which the system is able to provide a net energy return. Further discussion of the EPBT and EROI methodology can be found in the IEA PVPS Task 12 LCA Guidelines, and a discussion of its misrepresentation in a few publications is discussed by Raugei and colleagues.^{11, 12}

As shown in Figure 7.22, the EPBT of PVs has been reduced by almost two orders of magnitude over the last three decades, as material use, energy use, and efficiencies have been constantly improving. For example, the CED energy used in the life cycle of complete rooftop Si PV systems were 2700 and 2900 MJ/m², respectively, for multi‐ and monocrystalline Si modules, down from 5000 and 2700 in 2006. The EPBT of these systems is 1.8 years for module efficiencies of 13.2 and 14% correspondingly in rooftop installations under Southern European insolation of 1700 kWh/m²/year, with a grid efficiency (η_G) of 0.31 and a PR of 0.75 (Figure 7.23). The corresponding EROI ratio, assuming a 30‐year lifetime, is 17. In these estimates, the BOS for rooftop application accounts for 0.3 years of EPBT, but there are different types of rooftop mounting systems with different energy burdens.13

For CdTe PV, the primary energy consumption is 850 and 970 MJ/m², correspondingly based on actual production from First Solar’s plant in Frankfurt‐Oder, Germany, and Perrysburg, Ohio, United States. For insolation levels of 1700 kWh/m²/year and a grid efficiency (η_G) of 0.31 (values that correspond to average U.S. and South European conditions), the EPBT and EROI values for installed PV + BOS systems are EPBT = 0.65 years and EROI = 46 (Figure 7.25).

EPBT decreases (and EROI increases) as the solar irradiation levels increases; for example, in the southwest of the United States (latitude optimal irradiation of 2300 kWh/m²/year), the EPBTs of crystalline silicon and cadmium telluride PV in fixed‐tilt ground‐mount utility installations, respectively, are 1.2 and 0.5 years (Figure 7.26). In the highest solar irradiation regions (e.g., northern Chile; irradiation of 4000 kWh/m²/year) on a 1‐axis sun‐tracking plane, the EPBT for CdTe PV systems is only 3–4 months. Overall, PV systems are generating all the CED used in their life cycle in 0.4–2 years depending on the PV technology and the location where the system operates. With an assumed life expectancy of 30 years, their EROI is in the range of 15–90. So, depending mainly on the irradiation in the location where they are installed, PV systems generate 15–90 times more energy than the energy consumed in their life cycle, saving a tremendous amount of primary energy that can be used for building a new power infrastructure comprising PV modules and structures, power electronics, grid extension, and storage to enable the transition to a carbon‐free society.¹⁴

GHG emissions and global warming potential (GWP): The overall global warming potential (GWP) due to the emission of a number of GHGs along the various stages of the PV life cycle is typically estimated using an integrated time horizon of 100 years (GWP₁₀₀), whereby the following CO₂‐equivalent factors are used: 1 kg CH₄ = 23 kg CO₂‐eq, 1 kg N₂O = 296 kg CO₂‐eq, and 1 kg chlorofluorocarbons = 4 600–10 600 kg CO₂‐eq. It is noted that the CO₂‐equivalent factor of CH₄ has recently been updated to 32; the effect of this update on the reported GHG has not been assessed yet. Also if we consider a 20‐year time horizon, the GWP of CH₄ is a much higher 72. Electricity and fuel use during the production of the PV materials and modules are the main sources of the GHG emissions for PV cycles, and specifically the technologies and processes for generating the upstream electricity play an important role in determining the total GWP of PVs, since the higher the mixture of fossil fuels is in the grid, the higher are the GHG (and toxic) emissions.

In this chapter we present the most up‐to‐date estimates of EPBT, GHG emissions, and heavy metal emissions from the life cycles of the currently commercial PV technologies (e.g., monocrystalline Si, multicrystalline Si, and CdTe). The production of solar‐grade silicon and of silicon and thin‐film PV modules was discussed in Chapter 6. Life‐cycle inventory (LCI) data on materials, energy, and emissions of these technologies and of BOS can be found in IEA PVPS Task12 reports.

In the next paragraphs, we will briefly discuss the results of comparative assessments of various technologies.

The GHG emissions of Si modules are about 28 g CO₂‐eq./kWh range for a rooftop application under Southern European insolation of 1700 kWh/m²/year and a PR¹ of 0.75 (Figure 7.27). Notably, under these estimates, the BOS for rooftop applications account for approximately 5 g CO₂‐eq./kWh of GHG emissions.

The GHG emissions of CdTe PV systems total 14 g CO₂‐eq./kWh for irradiation levels of 1700 kWh/m²/year. The manufacturing of the module and upstream mining/smelting/purification operations comprises most of the energy and greenhouse burdens. The BOS share was about 5 g CO₂‐eq./kWh under 1700 kWh of insolation, with 14% module efficiency and a PR of 0.75. It is noted that the PV efficiencies shown in Figure 7.26 correspond to 2014 production and the current module efficiencies are significantly higher. As of mid‐2017, the average efficiencies of both CdTe and polycrystalline Si PV are approximately 16.5%; accordingly the GHG emissions during their lives would be 12 and 24 g CO₂‐eq./kWh of generated electricity.

All the recently published LCA studies found CdTe PV having the lowest EPBT and GHG emissions among the currently commercial PV technologies, as it uses less energy in its material processing and module manufacturing.⁹ This is explained by the lower thickness of the high purity semiconductor layer (i.e., 2–3 µm for CdTe vs. 15–200 µm for c‐Si) and by the fact that CdTe module processing has fewer steps and is much faster than c‐Si cell and module processing.

Figure 7.27 compares GHG emissions from the life cycle of PV with those of conventional fuel‐burning power plants, revealing the environmental advantage of using PV technologies. The majority of GHG emissions come from the operational stage for the coal, natural gas, and oil fuel cycles, while the material and device production accounts for nearly all the emissions for the PV cycles. With over 50% contributions, the GHG emissions from the electricity demand in the life cycle of PV are the most impactful input. Therefore, the LCA results strongly depend on the available electricity mix. In the United States replacing grid electricity with PV systems would result in an 89–98% reduction in the emissions of GHGs, criteria pollutants, heavy metals, and radioactive species.

However, this discussion is about large penetration of PVs, and so far we accounted for the GHG emission reductions enabled by PV displacing fossil fuel generation. For a balanced comparison, we also have to account for negative impacts resulting from the intermittency of the solar resource. Fluctuations in PV generation would necessitate fuel generators’ up‐ and down‐ramping and part‐load operation, causing additional emissions. The net emission savings from RE integration are due to the displaced conventional generation minus the incremental emissions from suboptimal and part‐load generation. Fthenakis and Nikolakakis modeled this impact in a scenario of adding PV and wind capacities in the west zone of the NYS grid totaling 1.2 GW in the winter and 2.7 MW in the summer and determined that the subsequent non‐optimum dispatch of natural gas peaker generators caused 5–8% higher emissions per kWh than when their output does not fluctuate to accommodate PV and wind penetration into the grid. More specifically the emissions of conventional generators are being raised due to RE integration from 376 to 406 g/kWh in the spring highest penetration case and from 380 to 397 g/kWh for the winter lowest penetration case. These increases are more than counterbalanced by the zero operational emissions of PV and wind.

Toxic gas emissions: The emissions of toxic gases (e.g., SO₂, NO_x) and heavy metals (e.g., As, Cd, Hg, Cr, Ni, Pb) during the life cycle of a PV system are largely proportional to the amount of fossil fuel consumed during its various phases, in particular processing and manufacturing PV materials. Figure 7.28 shows estimates of SO₂ and NO_x emissions. Heavy metals may be emitted directly from material processing and PV manufacturing and indirectly from generating the energy used at both stages. For the most part, they originate as trace metals in the coal used.

Heavy metal emissions: When considering heavy metal emissions in PV life cycles, it is important to distinguish between direct and indirect emissions. Indirect emissions are from electricity inputs during the life cycle of the PV system, while direct emissions are those released “on‐site” during mining, smelting, manufacturing, and the end‐of‐life disposal.

Electricity generated from CdTe PV has direct Cd emissions from raw material extraction and processing, module processing, and possible accidental releases of Cd if CdTe PV modules are exposed to fire. Both CdTe and crystalline silicon PV have indirect emissions from their fossil fuel‐based electricity inputs. For CdTe these were found to be 10 times bigger than the direct emissions showing the importance of the source of electricity in manufacturing solar PV systems. Other heavy metal emissions such as arsenic, chromium, lead, mercury, and nickel are emitted in trace amounts, also as a result of electricity and fuel inputs into the system. The heavy metal emissions from a coal‐fired power plant with state‐of‐the‐art air pollution control systems are 90–300 times bigger than that of solar electricity on a per unit of energy basis. The issue with cadmium bears special discussion as the element is used in CdTe/CdS and to a smaller extent in CIGS/CdS PVs. Direct emissions of cadmium in the life cycle of CdTe PV have been assessed in detail. They total 0.02 g/GWh of PV‐produced energy under an irradiation of 1700 kWh/m²/year; this includes emissions during fires on rooftop residential systems, quantified in experiments at BNL that simulated actual fires.¹⁵ These experiments were designed to replicate average conditions, and the estimated emissions were calculated by accounting for US fire statistics pointing to 1/10 000 houses catching fire over the course of a year in the United States where most houses have wood frames by assuming that all fires involve the roof. However, the indirect Cd emissions from electricity usage during the life cycle of CdTe PV modules (i.e., 0.23 g/GWh) are an order of magnitude greater than the direct ones (routine and accidental) (i.e., 0.02/GWh).² Cadmium emissions from the electricity demand for each module were assigned, assuming that the life‐cycle electricity for the silicon and CdTe PV modules was supplied by the Union for the Coordination of the Transmission of Energy’s (UCTE) (European) grid. The complete life‐cycle atmospheric Cd emissions, estimated by adding those from the usage of electricity and fuel in manufacturing and producing materials for various PV modules and BOS, were compared with the emissions from other electricity‐generating technologies (Figure 7.29).16 Undoubtedly, displacing the others with Cd PV markedly lowers the amount of Cd released into the air. Thus, every GWh of electricity generated by CdTe PV modules can prevent around 5 g of Cd air emissions if they are used instead of, or as a supplement to, the UCTE electricity grid. In addition, the direct emissions of Cd during the life cycle of CdTe PV are 10 times lower than the indirect ones due to use of electricity and fuel in the same life cycle and about 30 times less than those indirect emissions from crystalline PVs. The same applies to total (direct and indirect) emissions of other heavy metals (e.g., As, Cr, Pb, Hg, Ni); CdTe PV has the lowest CED and, consequently, the fewest heavy metal emission. Regardless of the particular PV technology, these emissions are extremely small compared with the emissions from the fossil fuel‐based plants that PV will replace. Furthermore, the external environmental costs of PVs are negligible in comparison with the external costs of fossil fuel life cycles.

Recycling would further improve the EPBT and the EROI of PV because recycling glass, aluminum, tellurium, and other semiconductor materials requires only a fraction of the energy consumed in their primary production, and, consequently, it produces only a fraction of GHG emissions.

7.3.2 Environmental Health and Safety (EHS) in PV Manufacturing

PV technologies have distinct environmental advantages for generating electricity over conventional technologies. The operation of PV systems does not produce any noise, toxic gas emissions, or GHGs. PV electricity generation, regardless of which technology is used, is a zero‐emission process. However, as with any energy source or product, there are EHS hazards associated with the manufacture of solar cells. The PV industry uses toxic and flammable substances, although in smaller amounts than many other industries, and use of hazardous chemicals can involve occupational and environmental hazards. Addressing EHS concerns was the focus of numerous studies of the National Photovoltaic EHS Assistance Center at the BNL, which operated till 2013 under the auspices of the US DOE. More than 150 articles highlighting these studies are posted in the Center’s website (www.bnl.gov/pv) and at the website of Columbia University’s Center for Life Cycle Analysis (www.clca.columbia.edu). This work has been done in cooperation with the US PV industry, which takes EHS issues very seriously and reacts proactively to concerns. The following text is a summary of EHS issues pertaining to the manufacture of crystalline‐Si (x‐Si), amorphous silicon (α‐Si), copper indium diselenide (CIS), CIGS, gallium arsenide (GaAs), and cadmium telluride (CdTe), which are currently commercially available.¹⁷

7.3.3 Recycling Programs

Recycling spent modules may not be seen as an immediate issue with developing solar energy. However, the rapid growth of solar energy eventually will result in problems of waste disposal within 20–30 years as end of life of PV would generate a significant amount of waste (about 100 tons/MW of decommissioned PV modules). Disposal of small quantities of PV modules in landfills should not cause environmental hazards when the modules pass the EPA leaching tests that are designed to simulate release conditions. However, it is widely recognized that in large scales of deployment and decommissioning, recycling end‐of‐life PV modules would be necessary to prevent risks of environmental pollution and to recover valuable materials. In Europe, the PV industry adopted a proactive approach that served them well during the transition to, a currently required, compliance with Waste Electrical and Electronic Equipment (WEEE) regulations. Germany’s Electrical and Electronic Equipment Act (ElectroG) requiring collection and recycling of electrical and electronic equipment (EEE) was extended to PV in mid‐2015 and is expected to become a global standard. The United States lacks a national policy and the necessary infrastructure to mandate PV recycling. Environmental regulations can determine the cost and complexity of dealing with end‐of‐life PV modules. If they were characterized as “hazardous,” then special requirements for material handling, disposal, record keeping, and reporting would escalate the cost of decommissioning modules.

Currently, there are well‐tested technical solutions (separation and material recovery processes) for c‐Si (wafer based) and CdTe PV products, but not for other technologies. The first step in recycling both types of modules is to separate the junction boxes, and for c‐Si the aluminum frames. Subsequent steps deal with separating the glass from the solar materials and metals. For c‐Si modules thermal treatment burns off the laminates to facilitate the separation processes (called module delamination). The most common way to achieve this is through pyrolysis, heating the module to 450–600°C to decompose the organic encapsulant. After delamination, the components are manually separated, the glass is sent to a glass recycling facility, and the silicon wafers are processed further, either by polishing and reusing the wafer or by recycling the silicon into a new wafer (Figure 7.30). From the separation steps, copper wire, aluminum frame, glass, silicon, and waste are separated and are sent to recyclers. Plastic is burned off during the thermal treatment and waste goes to a landfill.

First, the unloaded modules transported from the collection sites will be loaded to the automatic conveyor system to enter the recycling process. Then the junction boxes are removed manually. Thermal treatment burns off the laminates to facilitate the separation processes. From the separation steps, copper wire, aluminum frame, glass, and waste are separated. During the next step the solar cells are treated chemically. Surface and diffusion layers are removed subsequently by cleaning steps. Cells and wafer breakage are cleaned by etching techniques. Junction box is processed by an electronic scrap waste treatment company (i.e., collection cost paid by PVTBC). Plastic is burned off after the thermal treatment (i.e., incineration cost paid by PVTBC). Waste goes to a landfill and PVTBC pays the landfill tipping fees. Aluminum can be reused while glass, copper, and silicon can be sold to recycling companies. The thermal process could be improved regarding its throughput, cycle time, and yield. The yield of recovered cells depends largely on the type, design, and state of the modules to be processed. Design‐dependent factors that affect the results of the thermal process are type of laminate, crystal type and dimensions of the embedded cells, and the material and dimensions of bonds and soldering.

Recycling of CdTe PV is somewhat more advanced and is employed in all CdTe PV production facilities. It is based on a low‐cost, environmentally friendly hydrometallurgical technology, which was developed by First Solar and BNL. This technology involves crashing the modules, removing the thin films from the substrate, and recovering the thin‐film materials from the solution (Figure 7.31). The modules are cut by a shredder and broken in small pieces with a hammer mill. The pieces are then exposed to leaching using a dilute mixture of sulfuric acid and hydrogen peroxide, which extracts the metals (mostly copper) and semiconductor elements (tellurium and cadmium). Ion‐exchange column²⁵ is then used to separate the copper and the cadmium from the solution, resulting in a tellurium‐rich solution from where Te is then extracted by selective precipitation. Cadmium is rinsed out the column and is recovered electrolytically.²⁶ A modified version of this process is used by First Solar. Instead of separating the cadmium and tellurium using ion‐exchange columns, First Solar precipitates all of the metals and sends this sludge to a third party for further processing.

As we discussed in the beginning of this chapter, resource availability, affordability, and lowest possible environmental impacts are three major pillars of sustainable PV growth to levels that will enable transition of the current fossil fuel‐based electricity to a renewable one. Recycling of spent PV modules addresses all these three dimensions of sustainability (Figure 7.32). Recycling relieves the pressure on material prices as it creates an important secondary resource of materials and it eliminates the potential risks and liabilities associated with waste disposal. End‐of‐life PV contains materials of high value (e.g., silver, indium, tellurium, and gallium), toxic materials (e.g., cadmium lead, selenium compounds), and a large amount of energy‐intensive materials (e.g., glass, copper, steel, aluminum, and solar‐grade silicon). For these reasons, we must develop efficient, cost‐beneficial recycling processes to aid in creating a PV‐recycling economy that is essential to safeguarding and maintaining the SunShot‐catalyzed growth of the PV industry. The cost of solar energy is impacted by the price of material inputs such as polysilicon, tellurium, steel, and glass. Their prices are driven by the status of their supply and demand.

Recycling helps to avoid shortages of such materials needed for PV production and lowers the cost of PV modules. Currently recycling programs are established for only two types of PV modules: CdTe and c‐Si. The first recovers glass and the semiconductor elements for reuse in CdTe synthesis, whereas the second only recovers the aluminum frame and the glass. PV recycling of mature technologies (e.g., c‐Si and CdTe) is technically and economically feasible. Accounting for secondary production from recycling and the continuing improvements of module efficiencies and material utilization, a number of studies show that the availability of tellurium in the forthcoming decades is sufficient for a cumulative production at the TW level. In Europe, the PV industry established PV CYCLE, a voluntary program to recycle PV modules (www.pvcycle.org). This type of industry‐wide approach to economically manage large‐scale recycling should become an essential component of cost reduction roadmaps. Furthermore, we believe that a comparative techno‐economic evaluation of existing and proposed PV recycling technologies is necessary; this will aid the commercialization of cost‐beneficial recycling and the creation of a PV‐recycling economy.

Recycling of defective and end‐of‐life‐PV modules is a necessary component of the sustainability of the large‐scale deployment of PVs. There are many studies that urge a proactive systematic approach to resolving issues of the scarcity of materials and to assuring that PV life cycles will not present risks to public health nor the environment in scenarios involving a high penetration of PV in electricity grids. With approximately 16 GWs of PV already installed in the United States through 2014, over a million metric tons of PV module waste would be generated over the next 25 years in this country alone. Meeting the SunShot targets will cause a significant new waste source to the existing waste stream.

The concerns about PV waste are caused by issues associated with electronic waste (e‐waste). The Organization for Economic Co‐operation and Development (OECD) found e‐waste to be one the fastest‐growing waste streams in the Unites States. PV waste has the same characteristics as e‐waste, that is, it is a combination of valuable materials and toxic ones. It contains materials of high value (e.g., silver, indium, tellurium, and gallium), toxic materials (e.g., cadmium lead, selenium compounds), and a large amount of energy‐intensive materials (e.g., glass, copper, aluminum, and solar‐grade silicon). We examined the potential of PV in a prospective LCA, focusing on direct costs, resource availability, and environmental impacts, and showed that PV recycling will become increasingly important in resolving cost, resource, and environmental constraints to large scales of sustainable growth. Recycling helps to keep the cost down by creating a significant secondary source and resolves concerns regarding waste disposal (Figure 7.32). A major challenge is setting up cost‐effective collection infrastructure; this has been studied by Choi and Fthenakis.²⁷