When the U.S. space program began using them on its satellites in the late 1950s, solar cells cost more than $1,000 per peak watt (Wp). At that price, a photovoltaic system just large enough to keep a 100-watt bulb burning would cost more than $1 million.

While cell prices have fallen dramatically since then, to less than $10 per Wp today, solar photovoltaics still can't compete in cost with the electricity generated with fossil and nuclear fuels in utility power plants. Before solar cells begin to make economic sense to utilities or to individual homeowners, prices will have to fall to less than $1 per Wp. No one in the industry doubts that prices this low or lower are possible, but there is no consensus on when, or technically how, the industry will get there.

There are two ways to reduce the cost of electrical power produced by solar cells. More efficient cells would help.

Today's solar cells typically convert less than 15% of the sunlight that hits them into electricity. Efficiencies of 17% or 18% are possible with cells made from single crystals of silicon, the most widely used cell material, and more complex cells may someday reach efficiencies of almost 40%. But even they converted 100% of the incident sunlight to electricity, today's solar cells couldn't compete with power companies.

The only way to make solar-generated electricity competitive is to lower dramatically the cost of the cells, and that means lowering the cost of materials from which the cells are made.

First-generation photovoltaic cells, those commercially available today, are made, like most semiconductors, of silicon. The silicon has to be refined to remove impurities and then converted from a liquid to a crystal. Crystal growing, using the conventional Czochralski method, is time-consuming, and when the crystal is finally formed, it has to be sawed into thin wafers, which means that half the crystal ends up on the floor as expensive sawdust.

The wafers become solar cells when small amounts of boron and phosphorus are added to opposite sides, creating a semiconductor junction within the wafer. Electrical contacts are added, and, when the sun strikes one side of the cell, a small electrical current flows through a wire connecting the two sides. To double the current, the number of cells is doubled. The relationship between the power produced and the number of cells used is strictly linear. There are no economies of scale with solar cells.

Individual cells are only a few centimeters in diameter. A large number of them are usually connected electrically and sealed between glass or plastic into photovoltaic modules. In turn, any number of modules can be combined into a roof-top or free-standing array of solar cells.

The question is how to cut the cost. Manufacturers and researchers are pursuing several alternatives, none of which has yet proven superior.

Computer-controlled crystal-growing machines and high-speed diamond saws, along with cheaper refining techniques, continue to bring down the cost of the single-crystal silicon wafers. ARCO Solar Inc., the world's largest producer of single-crystal solar cells, recently agreed to supply a large number of these first-generation cells at $4.90 per Wp.

Other companies are making cells from noncrystalline silicon. Instead of going through the lengthy crystal-growing process, Solarex Corp., in Rockville, Md., pours molten silicon into a cube-shaped crucible, where it cools and solidifies. The resulting block, which is not a single crystal but a mixture of crystals, is sliced into rectangular wafers that make more efficient use of the module space than round wafers. Efficiencies are not yet as high in these polycrystalline cells as in the single-crystal variety.

Amorphous silicon, a thin-film of noncrystalline silicon applied to glass or another substrate, makes cells still less costly. Energy Conversion Devices Inc. of Troy, Mich., recently formed a joint venture with Japan's Sharp Corp. to begin automated production of amorphous silicon solar cells in Japan. The cells will power Sharp pocket calculators.

Other companies are looking into the use of entirely different materials for thin film solar cells -- cadmium sulfide, gallium arsenide, cadmium telluride, and others.

Before photovoltaic cells are ready for widespread use in large generating systems, most experts believe that cell efficiencies will have to be raised, cell prices will have to be cut drastically, and elaborate systems for concentrating sunlight on the cells will have to be devised. Every year the projected schedule for this kind of application gets pushed further back, and falling oil prices don't make major corporations any more eager to commit funds to photovoltaic development.