These tiny indentations, each less than a micrometer across, can trap rays of light as effectively as conventional solid silicon surfaces that are 30 times thicker, according to their findings. The new findings are reported by MIT postdoc Anastassios Mavrokefalos, professor Gang Chen, and three other postdocs and graduate students, all of MIT's Department of Mechanical Engineering. "We see our method as enhancing the performance of thin-film solar cells," Mavrokefalos said, but it would actually work for any silicon cells. "It would enhance the efficiency, no matter what the thickness".

Graduate student Matthew Branham, a co-author of the paper, said, "If you can dramatically cut the amount of silicon (in a solar cell) ... you can potentially make a big difference in the cost of production. The problem is, when you make it very thin, it doesn't absorb light as well."

The operation of a solar cell occurs in two basic steps: First, an incoming particle of light, called a photon, enters and is absorbed by the material, rather than reflecting off its surface or passing right through. Second, electrons knocked loose from their atoms when that photon is absorbed then need to make their way to a wire where they can be harnessed to produce an electrical current, rather than just being trapped by other atoms. Unfortunately, most efforts to increase the ability of thin crystalline silicon to trap photons, such as by creating a forest of tiny silicon nanowires on the surface, also greatly increase the material's surface area, increasing the chance that electrons will recombine on the surface before they can be harnessed.

The new approach avoids that problem. The tiny surface indentations described a "inverted nanopyramids" greatly increase light absorption, but with only a 70 percent increase in surface area, limiting surface recombination. Using this method, a sheet of crystalline silicon just 10 micrometers (millionths of a meter) thick can absorb light as efficiently as a conventional silicon wafer 30 times as thick.

That could not only reduce the amount of expensive, highly purified silicon needed to make the solar cells, Mavrokefalos explained, but also cut the weight of the cells, which in turn would reduce the material needed for frames and supports. The potential cost savings are "not only in the cell material, but also in the installation costs."

In addition, the technique developed by Mavrokefalos and his colleagues uses equipment and materials that are already standard parts of silicon-chip processing, so no new manufacturing machinery or chemicals would be required.

Top view of a sheet of silicon textured with an array of micrometer-scale inverted-pyramid shapes, for efficiently trapping light waves. Source: Anastassios Mavrokefalos, MIT

To create the tiny dents, the researchers use two sets of overlapping laser beams to produce exceptionally tiny holes in a layer of material — called a photoresist — that is deposited on top of the silicon. This interference lithography technique is scalable to a large area. Following several intermediate steps, a chemical called potassium hydroxide is used to dissolve away parts of the surface that were not covered by the photoresist. The crystal structure of silicon leads this etching process to produce the desired pyramidal shapes in the surface, Mavrokefalos said.

So far, the team has only carried out the first step toward making the new type of solar cells, producing the patterned surface on a silicon wafer and demonstrating its improvement in trapping light. The next step will be to add components to produce an actual photovoltaic cell and then show that its efficiency is comparable to that of conventional solar cells. The expectation is that the new approach should produce energy-conversion efficiencies of about 20 percent, compared to 24 percent for the best current commercial silicon solar cells, but this remains to be proved in practice.