Photovoltaics come down to earth8 July 2004
A personal account by the firm's founder of Solaicx's achievement of a low cost high efficiency PV wafer, set in the context of fifty years of solar cell development
Invented at Bell Labs 50 years ago, power via photovoltaics, or solar cells, has left its mark on a wide range of applications that most of us take for granted now. In its infancy, solar cell technology found its first economical use in remote applications like space satellites, then later stayed alive by playing in niche markets like remote offshore power and oil production and mountain-top relay stations. More than half a century has gone by and the world of photovoltaics has seen impressive strides, with the market growing at a rate of 35 % a year.
Solar cells in space
The PV power systems currently on the market have developed over the course of forty years, with the space industry being its very first frontier. For the first twenty years (1960 to 1980) solar cells were used to power spacecraft in orbit, in lieu of other possible power sources. These early years developed important attributes of the cell such as reliable contacts, encapsulants, optical properties and test standards. My first spacecraft job, at Boeing in 1965, was testing the Lunar Orbiter spacecraft, powered by solar cells from RCA Mountain Top in Pennsylvania. In the 1960s solar cells were just another diode on the production lines at RCA, TI, Motorola and Hoffman. In 1967 I left Boeing and became president of Spectrolab as America went to the moon. The Apollo 11 landing on the moon in 1969 left behind a Spectrolab solar array powering a seismic monitor. The solar powered unit remained operational for several years until NASA turned off the transmitter.
The cost race begins
The next ten years (1980 to 1990) saw an expansion of solar cells, from their use in space, which was a relatively small market, to a number of terrestrial applications including remote-power markets such as offshore oil platforms, pipeline instrumentation, remote landing-strip beacons, mountain-top communications repeaters, etc. Most of this market involved charging batteries in remote locations. Just like the spacecraft application of solar cells, remote power systems operate unattended. All through these early 30 years, the growth rate of the PV business was a healthy 25 % per year or more.
I started Solar Technology Inc in 1975 with one goal: To make much lower-cost solar panels for industrial use. My first solar panels produced 20 W for charging a 12-volt battery and sold for $300. So the cost metric of $15/W began to be used to compare terrestrial solar panels. Since then, the push has been to reduce cost but retain the reliability the space business taught us. I sold my company to Atlantic Richfield in 1978 and we renamed it ARCO Solar. Two years later, we finished building a new factory in Camarillo, CA, which still remains the largest plant in the USA. This plant mass-produced large, 35 W, panels that sold for $300 and reduced the price from $15/W to $8/W. In 1987 ARCO eventually sold the company to Siemens Solar, and in 2001 Siemens sold it to Royal Dutch Shell, renaming it Shell Solar. During this 24-year period production kept increasing and costs continued to drop on a predictable curve.
The utilities get on board
The latest period of rapid growth (1990 to 2004) has shifted from battery charging to grid-connected home applications in California, Germany and Japan. Called net-metering, the solar panels are connected to inverters which transfer DC power into the AC grid. During daytime the electric meter may actually run backwards, crediting the homeowner. At night the power systems provide the home with electricity from the grid. Ninety per cent of this business uses low cost versions of the original silicon (simple PN junction) cell used in spacecraft in the early days. The price for terrestrial solar electric modules (per peak W) has dropped from the spacecraft $1000/W to the current best price of $3/W. Warranties for commercial modules are 25 years and there are virtually no returns.
Three international companies that began making terrestrial solar panels in the early 1970’s included Shell Solar in California (formerly ARCO Solar), BP Solar in Maryland (formerly Solarex), and Sharp Solar in Nara, Japan. All of these companies have similar panels over 150 W and factory prices of around $3/W.
Manufacturing advances are not the only factor driving solar adoption. In recent years, Japan, Germany, and the world’s sixth-largest economy, California, have fashioned a variety of incentives and policy changes that have made them the centres of the budding solar industry. In California, two programmes have spurred solar adoption: a rebate that helps defray the cost of installing small-to-mid-sized solar arrays and a policy that allows solar users to sell electricity to the utilities (net metering). In Germany, “Green Party” subsidies are ramping up solar panel installations. The government will pay a subsidy for power generated by panels installed on the home. The weather in Germany, however, means there is only half the energy available annually compared to California installations. The current Japanese plan calls for maximum use of PV powered homes on the island nation, where it was energy concerns that led it into World War II. Sharp Corporation, Kyocera and Sanyo are the major players in Japan, pushing to expand their worldwide production. In Europe, BP Solar, Shell Solar and RWE/Schott are major players. All three of these companies have major solar cell plants in the US.
Since the invention of the silicon solar cell at Bell Laboratories in 1954 many changes have helped to continue bringing the cost down, and increasing volume. Our first terrestrial solar modules at Spectrolab, in 1975 were made with 2.5 inch diameter silicon wafers and were priced at $25/W. When I started Solar Technology Inc in 1975 I used
3 in diameter wafers and was able to reduce my price to $15/W and make a good profit. At ARCO Solar in 1980 we built a more automated production line, in Camarillo, CA, and used four-inch wafers to drop the price to $8/W. All of these wafers were single crystal, grown on semiconductor Czochralski (CZ) growers used for solar wafers. This trend has continued with ARCO Solar becoming Siemens Solar, and now becoming Shell Solar. Siemens moved the crystal grower plant from Camarillo, CA, to Camas, WA, near Portland to reduce the cost of electric power. There are now 69 converted CZ growers in that plant connected to the hydroelectric power from the Columbia River. The wafers in the current Shell Solar modules are six-inches in diameter and the volume is up to 80 000 wafers per day (24/7) reducing the module prices to around $3/W in container quantities. Other cost reductions in the module and cell design have used robots to make labour more productive. As the volume has increased, the costs have been reduced in a classic expansion of the business. Engineers out in the power system world have also found clever ways to use photovoltaic power modules to solve problems. In addition, the industry of solar cell and module production companies have shaken-out to about 10 large producers and 10 smaller ones around the world. Sharp Solar of Nara, Japan is the leader currently followed by BP Solar (ex-Solarex), Kyocera, Shell Solar (ex ARCO Solar), Sanyo (with a new 20% efficient cell), SunPower Corp. (with a new 21% efficient cell), Mitsubishi, AstroPower, etc.
Cost reductions with new technologies
The $4 billion per year business that has developed is still growing at 35% per year, and costs are still dropping on a predictable learning curve. The first cost element of silicon solar cells is the pure silicon. The semiconductor business used the original Siemens batch process to make large chunks of pure silicon. New continuous fluid bed reactors are now available to reduce cost and energy. These new reactors are beginning to replace the original Siemens slim rod reactors that used 90 kw-hours per kilogram of silicon produced. Fluid-bed reactors produce silicon for 15 kWh/kg, lowering the payback time for solar panels. Modern silicon solar-module processes can pay back the energy of manufacturing in as little as two years in California weather. This was of little interest to the semiconductor manufacturers.
The second step requires making the pure silicon into a wafer by melting it and then freezing it in a controlled way. Some processes make ribbons, some processes cast blocks and saw them up, and, as in the semiconductor business, the Czochralski (CZ) grower makes single crystal solar wafers with the highest efficiency. Silicon wafers, about 300 microns thick, absorb the sunlight and produce electrons and holes. The silicon wafers are then made into solar cells, which are just one big diode with contacts for plus and minus electrical connections. The current is directly proportional to the intensity of the solar energy. Currently available cells are made from six-inch-diameter CZ wafers and produce from two to three W each, depending on manufacturer. Each cell, in sunlight, produces about half a volt and five amps at maximum power. Other companies are beginning to use eight-inch-diameter wafers to reduce costs per W.
The third step in making a solar module is to interconnect 36 solar cells in series producing 20 volts for 12-volt battery charging. A solar module makes DC electricity and is current limited by the available sunlight. After the series of solar cells are connected, they are laminated behind a sheet of tempered glass similar to a windshield. The solar cells then look at the sun through the glass, which is washed clean when it rains. The front of all modern solar panels is tempered glass and automotive windshield lamination techniques are used for encapsulation. Automotive windshields have been tested in outdoor sunlight for millions of vehicles and more than 50 years. Usually the tempered glass modules are edge protected with aluminum extrusions. The extrusions and laminated glass panels are strong enough for workers to walk on them during installation. Solar panels are modular, so large or small power systems can therefore be made in the field. Panels are added in series and parallel. Thus a power system can be expanded later if the load grows or was underestimated. The local sunlight is always an estimate, bad weather might also require a solar array be enlarged at a later time.
Still, there are many more improvements that can be made to further reduce the costs of silicon solar panels. I mentioned the trend to move from the Siemens process silicon production using a batch process to the new “fluid bed” continuous production of silicon bb’s. The silicon bb’s are used by a number of solar ribbon growers and others, but don’t melt down properly in semiconductor CZ crystal growers.
Solaicx Inc (my new start-up company in Santa Clara, CA) is making a big improvement in the CZ crystal grower. Our new, patented, radically different crystal grower runs continuously, never shutting down the molten silicon. We use the new low cost fluid-bed silicon in our machine loaded automatically by computer control, not by hand by an operator in a clean room. Solaicx is dramatically lowering the cost of high minority-carrier lifetime wafers using a proprietary technology that reduces the cost of solar electricity. Solar electricity can finally be cost competitive with traditional forms of power generation. In California, with its high summer prices, this means in effect beating $0.25 per kWh.
As the PV business is growing several silicon manufacturers are starting up fluid-bed silicon production lines to increase supply for PV. The use of robots is reducing cost in the solar cell and module lines.
A PV market in the home
Panels for remote power are usually connected to batteries. Panels for grid connected housing (the largest current market) are connected to inverters that track the peak power of the panel and provide AC power. The most popular inverters are in the two to three kw size range and the solar panel DC input is about 200 volts. Currently costs are generally divided into three parts, one-third for the silicon wafer, one-third for the solar cell/module and one-third for installation (system). In California the market is growing because builders are starting to incorporate solar panels in each house in subdivisions. California is a summer air conditioning load-driven electric system. If you build 50 homes in California you add fifty air conditioners to the utility load. However if the houses have built in solar panels facing west the added subdivision summer afternoon electrical load is minimal. Also important is the current trend for these solar panels to be incorporated in the home loan. Solar panels all have a 25year standard warranty and are UL approved. The inclusion of solar modules in the home design and mortgage has really reduced the cost of installed PV power.
New solar modules are getting larger, with many 200 W units now available. This larger size and higher efficiency results in fewer trips up the ladder during installation of a given system. Installation is a major part of the cost. Solaicx helps bring down the downstream costs by as much as 50%. Design changes, plus building the modules into the house roof, results in more productivity, hence lower cost per kW installed.
Even the battery units used in remote sites are now often assembled with structure and wiring complete at the factory. Factory assembled systems are trucked to the field to be lifted into position with minimum field labor. Solar module assemblies are light in weight and can be flown to offshore oil platforms on small helicopters. 2 kW systems ready to install can be carried to a job site in the back of a pickup truck and carried onto the roof. The new 2 kW inverters from Fronius (Germany) can be carried by hand and easily installed on a wall.
A bright future
1980 was the first year any factory in the world produced and sold one megawatt in one year. That was the first year of start-up for the ARCO Solar Camarillo, CA plant. In the year 2000 I was invited back to Camarillo to celebrate the production of 200 MW of accumulated production since 1980. This year, 2004, Shell Solar is expanding production from 80 MW to 100 MW per year, which includes a new plant in Germany. Sharp is building new plants in Memphis, Tennessee and Wrexham, North Wales. In the next few years Sharp Solar projects reaching production of one GW per year from their present 300-MW level. I am proud of the progress our industry has made in providing useful electric power supplies at ever lower prices for industry use. The solar-electric-power business is accelerating rapidly into a bright future.