Ultra-Efficient Solar Power

Harry Atwater thinks his lab can make an affordable device that produces more than twice the solar power generated by today’s panels. The feat is possible, says the Caltech professor of materials science and applied physics, because of recent advances in the ability to manipulate light at a very small scale.

Solar panels on the market today consist of cells made from a single semiconducting material, usually silicon. Since the material absorbs only a narrow band of the solar spectrum, much of sunlight’s energy is lost as heat: these panels typically convert less than 20 percent of that energy into electricity. But the device that ­Atwater and his colleagues have in mind would have an efficiency of at least 50 percent. It would use a design that efficiently splits sunlight, as a prism does, into six to eight component wavelengths—each one of which produces a different color of light. Each color would then be dispersed to a cell made of a semiconductor that can absorb it.

Atwater’s team is working on three designs. In one (see illustration), for which the group has made a prototype, sunlight is collected by a reflective metal trough and directed at a specific angle into a structure made of a transparent insulating material. Coating the outside of the transparent structure are multiple solar cells, each made from one of six to eight different semiconductors. Once light enters the material, it encounters a series of thin optical filters. Each one allows a single color to pass through to illuminate a cell that can absorb it; the remaining colors are reflected toward other filters designed to let them through.

Another design would employ nanoscale optical filters that could filter light coming from all angles. And a third would use a hologram instead of filters to split the spectrum. While the designs are different, the basic idea is the same: combine conventionally designed cells with optical techniques to efficiently harness sunlight’s broad spectrum and waste much less of its energy.

It’s not yet clear which design will offer the best performance, says Atwater. But the devices envisioned would be less complex than many electronics on the market today, he says, which makes him confident that once a compelling prototype is fabricated and optimized, it could be commercialized in a practical way.

Achieving ultrahigh efficiency in solar designs should be a primary goal of the industry, argues Atwater, since it’s now “the best lever we have” for reducing the cost of solar power. That’s because prices for solar panels have plummeted over the past few years, so continuing to focus on making them less expensive would have little impact on the overall cost of a solar power system; expenses related to things like wiring, land, permitting, and labor now make up the vast majority of that cost. Making modules more efficient would mean that fewer panels would be needed to produce the same amount of power, so the costs of hardware and installation could be greatly reduced. “Within a few years,” Atwater says, “there won’t be any point to working on technology that has efficiency that’s less than 20 percent.”