PHOTOVOLTAIC TECHNOLOGY AND MARKETS

I. INTRODUCTION No form of "alternative" or renewable energy has received more attention than photovoltaics. Literally billions of dollars have been spent researching materials and processes that directly convert sunlight to electricity. The first "solar cell" was made in the Bell Laboratories in the early days of the semiconductor revolution--about the same time as the first transistor. Clearly, the transistor already has revolutionized modern life. Yet people still talk about the future "photovoltaic revolution." Early commercial applications of photovoltaics still are operating some 25 years after installation, and hundreds of thousands of installations have been made. However, the sun's major promise as an energy source is yet to be felt. Nevertheless, public policy interest, government funding, and, most importantly, private risk capital continue pouring into this promising technology. The rationale for public support has its roots in the oil crisis of the early 1970s. Since that time, the motivation for public support, which began as a fear of imminent shortage of fossil fuels, has been broadened to include affordable rural electrification in the third world, environmental protection, and industrial policy for high technology exports. This money and effort have created a relatively small but worldwide business poised on the brink of a commercial breakthrough. This paper explores the current state of the market and technology for photovoltaics and draws inferences for those who are attempting to push the industry over the brink into greatness. II. PHOTOVOLTAIC OVERVIEW Energy reaching the Earth's surface from the sun is measured in units of kilowatt hours per day per square meter (kwh/d/|m.sup.2^). The average amount of sunlight at any location depends on latitude, cloud cover, and atmospheric conditions such as humidity. The distribution of this energy is more uniform than commonly believed. The Olympic rain forest in Washington receives about 3 kwh/d/|m.sup.2^ while southern Arizona, New Mexico, and parts of Utah receive about 7 kwh/d/|m.sup.2^. A more common way to express this measurement is to give the equivalent hours per day of "full sunlight." "Full sunlight" is defined as 1,000 watts per square meter of global radiation. Therefore, Arizona at 7 kwh/square meter gets seven hours of full sun while the Olympic rain forest at 3 kwh/square meter gets three hours of full sun. Reasonable historical insolation data are available for almost any location on Earth. Most photovoltaic companies predict their products' output using computer models that include these insolation data. The rule of thumb for predicting an unknown site is plus or minus 10 percent on an annual average based only on latitude and closest weather data. Well studied sites can be predicted within about 3 percent of annual output. Photovoltaic installations do not have much economy of scale. The cost per kilowatt drops very little above a site size of about 25 kw. Construction costs are related to the rate of installation, not the size of each segment. A construction site resembles a "factory in the field." Construction can be very quick. The ARCO Solar Carrisa Plains 6.5 MW site was built in five months, and techniques used on that site could reach 25 MW per month. Additionally, power can be turned on in increments as each segment of about 1 MW is completed. Therefore, there is almost no interest during construction and no start up period for a purchaser. Even large sites are unmanned, and there are no consumables. Published estimates of 0.8 cents per kwh grossly overestimate operation and maintenance costs. These estimates invariably come from small demonstration sites where maintenance of the data acquisition system and plant tours comprise the bulk of technician time. In summary, the resource is free and ubiquitous. Plant siting is easy, construction is quick, and performance is reliable and predictable. …