X

Silicon vs. CIGS: With solar energy, the issue is material

What should solar panels be made of? Silicon has history on its side, but the future may lie in CIGS.

Michael Kanellos Staff Writer, CNET News.com
Michael Kanellos is editor at large at CNET News.com, where he covers hardware, research and development, start-ups and the tech industry overseas.
Michael Kanellos
7 min read
Silicon or CIGS. In the solar world, them's fightin' words.

The booming solar industry is in the midst of an argument over which material will become dominant in the future for harvesting sunlight and turning it into electricity. Solar panels made from crystalline silicon currently account for more than 90 percent of the solar infrastructure today.

Unfortunately, silicon panels remain relatively expensive to make. Without subsidies, it's still cheaper to get electricity from the grid. A two-year shortage of polysilicon, which may not ease until 2008, has severely limited growth and sales.

Panels that harvest energy with CIGS (copper indium gallium selenide) cost far less to make and install, say backers. The material can be sprayed onto foil, plastic or glass or incorporated into cement and other building materials. Conceivably, the entire exterior of a house or building could become a solar generator.

CIGS also doesn't degrade in sunlight like other thin-film technologies.

"The smartest investors are going short on silicon and long on thin film, especially CIGS," said Martin Roscheisen, CEO of Nanosolar, a start-up that has received $100 million in venture funds to build a plant capable of producing 430 megawatts-worth of CIGS panels.

Silicon solar panels

"The semiconductor is 100 times thinner. We combine low-cost materials with low-cost processes. The expenses on silicon are extremely high."

A huge vote of confidence in CIGS came earlier this year when Shell, one of the largest solar companies in the world, sold its silicon solar business to focus on developing CIGS.

So if CIGS is so good, why isn't there more of it out there? Mind share.

Silicon has become one of the most studied materials ever discovered, and advances in reducing processing time and manufacturing that were discovered in the semiconductor world rebound directly to silicon solar-cell manufacturers. Other alternatives--solar thermal energy, photovoltaic dyes--have failed to undercut it in functionality and cost.

"Silicon has a reliability record which is unmatched by any other material," said T.J. Rodgers, CEO of Cypress Semiconductor, which is the primary stockholder in the fast-growing silicon panel maker SunPower.

"They could rename the company NanoDollar, because that's all they are going to be left with after we get done kicking their butt," Rodgers said referring to Nanosolar.

"The smartest investors are going short on silicon and long on thin film, especially CIGS."
--Martin Roscheisen, CEO of Nanosolar

He's got a point. Back in the early 1990s, CIGS was emerging as an alternative to silicon, but the declining price of silicon snuffed out the movement.

"The three most studied materials in history are steel, cement and silicon, so they have a leg up on us there," acknowledged B.J. Stanberry, CEO of CIGS developer HelioVolt. "I'd say you're a fool if you predicted the imminent death of silicon. But their inability to deliver is creating an opportunity for thin film, and CIGS will have a significant portion of the market within 10 years."

With demand cranking up to an all-time high for solar technology, the two types of panels will likely co-exist for years--especially considering the miniscule role solar plays now in generating electricity, according to various estimates, and that demand is expected to double by 2025. Solar accounts for less than 0.10 percent of the current total.

Nonetheless, growing momentum for one technology among researchers, equipment makers and, ultimately, customers could pave the way for one to become dominant over the other.

Similar debates weighing promise against pragmatism have occurred in chipmaking. Gallium, indium and germanium have also been used to produce superfast semiconductors, but the higher costs associated with these materials have kept them toward the margins in the market.

Silicon hits and misses

Silicon, even its adherents admit, is not ideal. Theoretically, silicon is capable of converting 29 percent of the sunlight that strikes it into electricity, according to Dick Swanson, a former Stanford professor who founded SunPower.

"That imagines a cell that is perfect in every possible way. That would be without any energy losses or leaks other than those demanded by the physics of silicon," Swanson said. "The practical limit, most say it is around 25 percent to 26 percent."

SunPower already sells panels that convert an average of 20 percent of the sunlight into electricity and will come out later this year with panels that will convert 22 percent. The high efficiency is due to the design of the company's panels. SunPower puts the electrical contacts at the back (or bottom) of the panel to increase surface area. The silicon also sits atop a reflective layer: Photos that would otherwise pass through the panel entirely are bounced back into it and effectively recycled.

Most other solar makers sit at 15 percent to 18 percent efficiency. Still, a physical limit is a physical limit and silicon makers acknowledge they are approaching a barrier. Additional layers made of different materials could be added to silicon panels to harvest more energy, but that adds to the cost.

Progress in the industry instead revolves around reducing the cost of the panels. So far, it's working. SunPower, among others, has figured out ways to automate many factory procedures. It also builds factories in the Philippines, where labor remains cheap. Panels are also getting thinner, which reduces the material needed and increases efficiency.

"Silicon has a reliability record which is unmatched by any other material."
--T.J. Rodgers, CEO of Cypress Semiconductor

Right now, it takes about nine to 10 years for the cost of a solar installation to pay for itself--meaning the cost equals the amount you would have paid the power company in electric bills. In five years, silicon makers claim they can cut that time in half.

And as an added bonus, solar panels aren't as ugly as they used to be. PowerLight has come out with roof tiles with embedded silicon solar panels, which get installed when a house is built. A complete system can run around $8,000 to $13,000, according to Grupe Homes, which has included PowerLight panels in some homes in a few relatively new developments.

Solar needs real estate

The problem, however, is that solar electricity takes a lot of real estate, said Stanberry. The sun radiates about a kilowatt of energy per square meter on the surface of earth. There are 2.6 million square meters in a square mile. Thus, every square mile gets about 2.6 gigawatts. (A million kilowatts equals a gigawatt.)

On a practical level, solar energy is only going to harvest about 10 percent of the energy that hits a large area, so it takes about 4 square miles of solar panels to generate a gigawatt, or about the same amount of electricity provided by two power plants.

"If you look at the thousands of things that humans do, there are only three things that take up thousands of square miles: agriculture, highways and construction," said Stanberry. "The unavoidable goal of solar technology is how do you cover thousands of square miles inexpensively."

CIGS, say advocates, can do this because the panels are cheap to make. David Pearce, CEO of CIGS manufacturer Miasole, says that his company can erect a factory that can put out 100 megawatts worth of solar panels a year for $25 million. (The measurement means that, if you gathered all of the panels produced by the factory, they could provide 100 megawatts of power at the same time.)

Evergreen Solar, a silicon maker, plunked down $75 million to build a 30-megawatt facility in Germany in 2006. While extra capacity can be added more cheaply than the first 30 megawatts, CIGS still has a cost advantage, says Pearce.

"The battle is going to be won on the manufacturing floor. What we have to do is transfer this into high-volume production," he said at a recent conference.

In 2010, the costs of generating a watt of electricity from a CIGS panel--including installation and other expenses--will come to around $2.50, when you consider the lifetime of the panel. That will be roughly equal to grid power at the time, Pearce said.

By the end of next year, Miasole expects to have the installed capacity to produce 200 megawatts-worth of panels a year. Pearce further added that the company will be profitable next year.

While it trails in efficiency, CIGS is not far away, Miasole has shown a CIGS solar cell that converts 19.5 percent of sunlight into electricity; in a manufacturing environment, that means 15 percent to 16 percent efficiency, the company acknowledges.

So why isn't CIGS a perfect solution? It barely exists commercially and the alternatives don't have a great track record. Thirty years ago, producing solar energy cost about $100 a watt, said Swanson, so the U.S. Department of Energy began to fund alternatives to silicon. Now, it's about $8 or $9 and going down. The alternatives are just getting out of the research phase. Reliability of silicon, he added, is unquestioned.

The fact that CIGS can go on a variety of surfaces also may not be as big an advantage as it looks.

"There is a lot of roof space on American homes," said Ron Kenedi, general manager of the solar unit at Sharp, one of the big silicon solar makers.

Ultimately, the two technologies could co-exist by going into different applications, said Walter Nasdeo, an analyst at Ardour Capital.

"It's a hard question to answer. If you are talking about solar on a house, you're probably better off using silicon, particularly in the near term. It's been around for a long time," he said. By contrast, CIGS might be best suited for large industrial roofs or signs. Then over time, CIGS could build out a network of home installers.

"Right now what they (CIGS makers) face are engineering issues, not technical issues," he said.