Cellular Engineering

Mice with Sickle Cell Genes

Online Health Survey

Smog NOx Detector

Cleaning Up Dirty Silicon

Priority Service for the Internet

Climate Modeling Tools for Weather Forecasting

Imaging Liquids

Magic Sizes and Microscopic Crucibles

An x-ray fluorescence microprobe reveals nickel contamination in a polycrystalline silicon solar cell; copper was also present. Both contaminants were removed by manipulating the heating process during manufacturing.

Cleaning Up Dirty Silicon

Scott McHugo, a postdoctoral fellow at the Advanced Light Source (ALS), has found important clues to upgrading the performance of cells made from polycrystalline silicon. While common manufacturing methods yield cells that can convert sunlight to electricity with an efficiency of from 12 to 15 percent, improvements could push this to 19 percent, greatly increasing their commercial viability.

"When sunlight is absorbed by the cell, it frees electrons which start migrating in a random-walk fashion toward a junction between p-type silicon and an n-type layer such as diffused-in phosphorus," explains McHugo, who became interested in solar cells when he was doing graduate work in materials science at UC Berkeley. "If the electrons make it, they contribute to the cell's output. Often, however, before they reach the junction they recombine at specific sites in the crystal."

Researchers have previously shown that sites of high electronic recombination occur not primarily at grain boundaries in the polycrystalline material, but more often at dislocations in the crystal. McHugo was the first to show that these dislocations were "decorated" with metal impurities.

Iron was the first impurity McHugo spotted, but when he came to the ALS, he examined the tiny dislocations using the x-ray fluorescence microprobe beamline, built and operated by the Center for X-Ray Optics; by aligning the resulting spectra with maps of the defects made with a scanning electron microscope, he could compare defects and impurities directly. "That's when I found that not only iron but copper and nickel were also concentrated in these high-recombination sites."

Lately, McHugo's results indicate that the chemical state of the iron is not a silicide but an oxide. "As an oxide, the iron is highly stable and very difficult to remove." Solar cells are grown from molten silicon, cut into wafers, and finished by adding dopants and attaching contacts. Metal from valves, couplings, and other manufacturing machinery can contaminate the polycrystalline wafers. While single-crystal silicon would be easy to purify, it is too costly for an industry with a narrow profit margin; so is rigorous cleanliness.

Luckily, cleanliness is not the only path to purity. Doping wafers with phosphorus and sintering aluminum contacts onto them (which requires heating them almost to melting) both help in "gettering" the silicon — getting out the contaminants chemically. By adjusting time and temperature, these standard processes could be optimized to do a better job.

McHugo has shown that briefly annealing the finished cell at high temperatures is enough to remove copper and nickel precipitates of moderate size, and he is investigating techniques to remove stubborn iron impurities. "The task of improving the material after crystal growth is more difficult than originally envisioned," McHugo says, "since most researchers expected the iron to be a silicide, which is much easier to remove" than the oxide form. "We're looking at a two-step process, first subjecting the wafer to very high temperatures and then lowering the temperature to finish the processing of the solar cell."

Nevertheless, "Investigators have already achieved 19 percent efficiency in the lab with small samples; the challenge is to do it on the production line with full-sized solar cell wafers," says McHugo. "If even a dirty manufacturing run produces cells of 12 percent efficiency, and a manufacturer can make money at 15 percent, think how profitable cells of 19 percent would be."

Next: New Technology Demonstrates Priority Internet Service

Research Review Fall '98 Index | Berkeley Lab