Although solar cell technology for converting the sun's power into electricity has improved steadily in recent years, high costs and inefficiencies have kept it from being a serious replacement for fossil fuels. A few high-tech heavyweights—IBM, Intel and Hewlett-Packard (HP)—hope to change this using the same formula of mass production and commoditization that helped them make personal computers mainstream over the past three decades.

IBM last week announced plans to make solar panels covered with a thin film of chemical compounds. The idea is that the film, when applied to different surfaces such as glass or brick, can produce solar energy more efficiently than conventional silicon wafer–based solar cells—which are made of materials similar to those used to fabricate computer chips. (That's right—a company built on chips based on silicon is trying to get the world to move away from using it in solar cells.)

Also last week, Intel spun off a new solar tech company called SpectraWatt, which was born with $50 million in investment capital from Intel, Cogentrix Energy LLC, PCG Clean Energy and Technology Fund and Solon AG.

Meanwhile, HP earlier this month began licensing technology to Xtreme Energetics, Inc., in Livermore, Calif., designed to help that start-up company deliver rooftop solar energy systems that produce twice as much energy as conventional solar panels at half the cost.

The timing of the push is right, according to a June McKinsey and Company report: Although the cost of manufacturing and installing a photovoltaic solar power system has dropped over the past 20 years, the cost of generating electricity from conventional sources has been rising, along with the price of natural gas.

The McKinsey study noted that as competition grows in the market to deliver solar cells, competitors must relentlessly cut costs by improving manufacturing processes, investing in research and development, and moving production to low-cost countries. IBM, Intel and HP are well-positioned to leap these hurdles thanks to decades of fine-tuning manufacturing processes for PCs, computer processors, and other tech equipment, not to mention their multimillion dollar R&D budgets and worldwide facilities in places such as China and India.

This helps these companies improve the performance of their products while cutting costs. "We need [companies such as] IBM making this technology to get this stuff to the marketplace," says Larry Kazmerski, director of the National Center for Photovoltaics at the U.S. Department of Energy's National Renewable Energy Laboratory (NREL) in Golden, Colo.

A typical solar cell has an efficiency of 15 percent—about one sixth of the sunlight striking the cell generates electricity, according to NREL. Silicon wafer–based solar cells, which account for 90 percent of installed solar capacity, are more efficient, converting up to 20 percent of the sun's energy, whereas thin-film cells generally convert only about 10 percent. This gap is expected to close over the next few years, making thin-film cells—which cost half that of silicon-wafer cells—a better bet.

The least expensive solar cells today are made using thin films of cadmium telluride. The typical cadmium telluride cell converts solar energy into electricity with 10 percent efficiency at a cost of about $4 per watt—although NREL claims to have developed a prototype that converts at 16.5 percent efficiency. IBM's CIGS technology—for copper, indium, gallium and selenide—is a thin-film approach that will initially be able to convert at least 15 percent of the solar energy it receives into electricity, says Supratik Guha, a senior manager at IBM's T. J. Watson Research Center in Yorktown Heights, N.Y. That compares with 19.9 percent efficiency for NREL's version of CIGS thin-film solar cells.

Photovoltaic solar panels made from crystalline silicon can produce energy for between $2.25 and $3 per watt. CIGS on glass panels can drop that cost to $1.50 per watt, Guha says, but he also acknowledges that solar energy will not be able to compete with fossil fuel–based electricity until it costs less than $1 per watt.

One way IBM is hoping to cut costs is by making its cells through a process that dissolves the semiconductor materials in a solvent to create a solution that is placed on a glass or metal sheet and heated until the materials adhere. Until now, thin-film solar cells have been made in vacuum chambers—the semiconductor materials are placed on the glass or metal and then air is pumped out of the chamber, creating pressure that fuses the materials. But such vacuum chambers are expensive to build and complicated to operate (requiring monitoring of pressure, temperature, current, voltage, time and other factors during the process of making the thin film). Guha is cautiously optimistic that the new process will be an improvement, "Conceptually, this is simple," he says. "But it's still a largely unproven process."

IBM will be partnering with Tokyo Ohka Kogyo Co., Ltd.—a manufacturer of computer chips, LCD televisions and other devices that require surfaces treated with chemicals to develop the cells; he hopes to bring them to market by 2011.

Intel's SpectraWatt will start building a manufacturing facility in Oregon by the end of the year and plans to begin shipping thin-film solar cells made from crystalline silicon to solar panel makers by the middle of 2009. "All of the different solar technologies out there are all driving toward the same goal," says SpectraWatt chief executive Andrew Wilson, "finding a technology that creates a commodity out of solar energy. The customer cares about their cost per kilowatt hour."

Intel is looking at components of solar energy systems outside of the cell itself, says Wilson, who is tight-lipped about the specifics of SpectraWatt's technology. He points out that the solar cell is only about 25 percent of the technology required to make an energy collection and conversion system. "Intel is looking to drive [down] the cost of the overall system," he adds, "not just the solar cell."

HP and researchers at Oregon State University have been developing see-through transistors since 2002 by baking zinc tin oxide—a semiconductor—into glass or plastic . HP originally envisioned the technology as a way to create computer displays in car or train windows, but Xtreme Energetics wants by 2010 to begin including this circuitry in its solar panels to improve their ability gather sunlight—another take on the thin-film model.

Because the solar panels would be transparent, Xtreme is also hoping that architects and builders will be willing to incorporate these unobtrusive solar panels into their designs. "Today, most of these solar cell farms sit out in the desert," says Xtreme CEO Colin Williams. "The problem with these desert-based utilities is that they're not anywhere near their customers, so they have to transport the power, which leads to a loss of energy. We wanted to offer something that a company could install on the roof of its data center."

It is still much too early to determine whether IBM, Intel and HP will be as successful creating a mass market for solar energy on par with what they did for computers. "Thin film will be able to take the cost of electricity down to better compete with existing utility prices," says Vasilis Fthenakis, senior scientist at Brookhaven National Laboratory's National Photovoltaic Environment Research Center in Upton, N.Y., and a Columbia University professor of environmental engineering. "But we really need to see in detail what these companies are proposing and whether it will truly be different from the solar technology already being developed for the past several years."

The companies are funding their own research, without government subsidies or tax incentives, which may or may not come once the technology is ready to hit the market. "The big question," Fthenakis says, "is who can bring the price down in an unsubsidized market."