Gallium arsenide (GaAs) may soon be the new frontier in solar cells as they could have efficiencies twice those of silicon.
Costs for the material are high, and that has limited its usage, but two research groups have come up with ways to get much more out of GaAs.
Both teams figured out how to make extremely thin layers of GaAs. Harry Atwater’s group at Caltech developed a process that allows them to peel hundreds of thin layers off a large aggregate of the material, much like individual graphene sheets can peel off a block of graphite. The end result is an extremely thin film of GaAs.
John Rodgers, who works at the University of Illinois at Urbana-Champaign, grows thin layers of GaAs separated by a thin sacrificial layer. When the sacrificial layer etches away, you get a collection of thin GaAs chips; they can then recycle the silicon wafer, cutting down on the costs significantly. A plastic stamp can then pick up the chips and “print” them onto just about any surface, including one pre-patterned with wiring.
In the rare cases where GaAs chips see use here on Earth, they typically work in a concentrated solar system, where lenses pump as many photons into the chips as they can manage without melting. But these tracking and focusing systems add significantly to the cost of these systems. Both groups are thinking of doing some focusing, but going about it in different ways.
Rodgers, who can print large arrays of tiny GaAs chips, is managing costs by keeping things simple: His team’s process involves dropping a plastic sphere that acts as a lens on top of the chip. There are some ideas about how to manufacture more specialized spheres that focus the light more efficiently, but, for now, simplicity is the selling point.
Atwater is making the focusing device a more central part of his system for reasons that focused on the physics of what happens inside the chips. When a photon absorbs, it creates a free electron and a positively charged “hole.” There are three things that can happen to this pair. One is that they end up at electrodes, producing a useful current. One is they recombine uselessly, releasing the energy as heat. The third is they recombine by releasing another photon.
For Atwater, the key to an efficient photovoltaic material is minimizing the wasteful recombination of electrons and holes. And that means getting them to re-emit a photon — in his view, a good photovoltaic material is also a good LED. In these materials, photons end up absorbed and re-emitted a hundred times before being productively harvested, and the inside of the material is a sea of photons.
The danger here is that some of the photons escape back out of the material. To limit that, the GaAs can go on a reflective backing that sends the stray photons right back into the chip. The front has to let sunlight in, but then keep photons from escaping. To do that, he’s testing a system that looks a bit like two U’s with their bottoms fused (technically, it’s back-to-back parabolic lenses connected by a narrow aperture). This takes photons from a broad area and funnels them into the PV chip. The other end of the U acts like a reflective cap, making it very hard for a photon to escape from the chip without reflecting back into it.
At this point, researchers are pumping lots of photons into the GaAs device and keeping them there until they end up absorbed. The next way to up the efficiency is to have multiple independent photovoltaic devices, each tuned to different wavelengths. Traditionally, this occurred with layered devices, where each layer takes out a specific chunk of the spectrum.
And that’s what Rodgers’ team is already doing, by placing a standard triple-junction cell (which absorbs three different chunks of the spectrum) on top of a fourth cell that grabs yet another chunk. Atwater hopes to use the optic devices his team is working on to split the light into colors and direct them to independent photovoltaic devices.
Both of these faculty members have started companies to try to commercialize their work, and the devices they’re making are already above the 40 percent efficiency mark, which is double that of silicon cells. Rodgers’ company already has a 5MW capacity plant, and he said that scaling up production to an 80MW capacity plant would let them produce devices that are cost-competitive with coal.