72 Different Kinds of Solar Cells?

If you think about it, Intermolecular's prototyping system for solar cells works like a time machine.

Instead of running a test wafer with 20 identical cells, Intermolecular's system – a complex combination of semiconductor manufacturing equipment and software – lets researchers bake 72 different types of cells on the same wafer at once. On the system that is represented as 18 different formulas each tested on four different absorber layers. These different cells are then tested for a variety of desired characteristics.

Ideally, the technique compresses the time required to test new combinations of materials for advances in efficiency or performance, argues Craig Hunter, vice president of Intermolecular's solar group.

"How do you optimize production? This is the fundamental issue" in solar, he said. "There is a trend to get that extra 1 to 2 percent with a complex device."

In one prototyping test with a chip maker, Intermolecular managed to concoct 4,000 short loop samples (i.e., samples designed to highlight one or two features) in 18 months. Conventional processes yielded only 29. The company also prototypes for memory makers and will likely move into electrochromatic windows and other markets.

Elegant lab tricks like this are increasingly becoming the path for startups to make it in solar as manufacturers seek ways to boost performance and new companies (and their investors) try to figure out ways to pull in revenue without spending too much on factories.

The importance of efficiency waxes and wanes in the solar industry. Cost, after all, is the ultimate arbiter. From 2003 to 2008, manufacturers sought to lower costs by building factories in China or figuring out ways to make panels with "dirty" silicon. Efficiency was downplayed by many in part because of the cost and uncertainties surrounding any R&D project. But now with silicon in easy supply and a glut of factory capacity lingering worldwide, it's back in vogue because it can increase overall factory output virtually and give a manufacturer a potentially defensible edge.

"Efficiency is very important," says Dr. Andreas Bett, the deputy director of the Fraunhofer Institute for Solar Energy Systems.

Suntech Power Holdings is putting increased emphasis on efficiency with its Pluto modules. 1366 Technologies, which couldn't find the money to build a factory, is moving into the manufacturing tools and intellectual property market, similar to how Intermolecular operates. Innovalight is providing intellectual heft to JA Solar, a mass manufacturer.

Intermolecular, backed by CMEA among others, doesn't fit neatly into categories, which again appears to be an emerging trend. The core of the company's technology is a series of semiconductor manufacturing machines for depositing chemicals onto substrates: the Tempus F-10 through F-30. In conventional chemical deposition systems, chemicals are sputtered onto a substrate and the results are examined.

In Intermolecular's system, the chamber contains four sputtering guns with tightly constrained targets. The wafers also rotate the axis. In the end, this leads to arrays of different mixtures cooked at the same time.

"The only thing that is constant is the temperature," he said.

If you have a mental picture of those arrays of biological samples seen in biotechnology labs, that's not a coincidence. The massively parallel techniques for creating samples touted by Intermolecular and Wildcat Discovery Technologies for alternative energy companies come from the bio world. The big difference with the bio world is that pharmaceuticals companies are looking for new materials. In solar panels and chips, the main question is finding optimal combinations of existing materials.

"In semiconductors, they are more concerned about interaction," he said. "If it is a thin film and a deposited layer, we can work with it."

The company, though, does not license the technology per se or sell the equipment. Instead, it will engage with a customer to develop a particular advance in a cell or a chip. Intermolecular then obtains a royalty from sales of the end product.

Why not sell equipment? Hunter, among other execs, hails from equipment-dynamo Applied Materials. The factory equipment business is notoriously difficult: Participants have to raise massive amounts of capital and the potential pool of customers is relatively small. Serving as a straight R&D consultant, meanwhile, doesn't scale well. Conventional intellectual property licensing – where a company licenses a somewhat basic design or process to many manufacturers for royalties – is also difficult. In order to land contracts and "prove" their IP has value, IP companies typically have to sue a large company and win in court.

"We want some skin in the game," he said.

So far, no solar manufacturers have products on the market that were created via Intermolecular's techniques. However, the company is in negotiations with about six companies. A manufacturer could likely go from prototyping an anti-reflective coating to having something on the market devised via this technique in 12 to 18 months or less, he added. A full cell might take two to three years.

Photo of the Tempus F-30 via Intermolecular