The discovery came after the researchers noticed that the density changes depended on the type of surfactant they used, which led them to try combining multiple surfactants. They found that the right combinations could be used to exaggerate density disparities between nanotubes of different diameters, and also, surprisingly, disparities between semiconducting and metallic nanotubes. While they still have not identified the precise mechanism involved, the researchers believe it has to do with the marked difference in the ability of the semiconducting and metallic nanotube to be electrically polarized.

Andrew Rinzler, professor of physics at the University of Florida at Gainesville, says the method has produced "the best data I've seen so far." The resulting batches are pure enough, he says, for high-performance electronics. Indeed, Hersam says, the method can produce batches of semiconducting or metallic nanotubes with better than 99 percent purity.

What's more, the method has the potential to be scaled up for large-scale manufacturing. While the Northwestern researchers used lab-scale centrifuges capable of producing only milligrams of sorted nanotubes a day, Hersam says an industrial-scale centrifuge could purify 100,000 times as much. Large-scale manufacturing, he says, would involve running many of these centrifuges in parallel, twenty-four hours a day, with products removed and new batches added while the machines continue to spin.

Rinzler and Martel caution that the effectiveness of the method at a larger scale still needs to be demonstrated. Furthermore, Martel says, tests are needed to demonstrate that individual nanotubes are not damaged during the process. However, since surfactants don't typically cause damage to materials such as carbon nanotubes, he doesn't expect that this will be a problem.

Indeed, researchers are already making use of the new technique. In the Nature Nanotechnology paper, Hersam reports employing the method to make relatively simple transistors using thin-film meshes of semiconducting nanotubes. Such transistors could be useful for controlling pixels in computer monitors and flat-screen TVs.

At the same time, Martel is using the sorting technique to help produce transparent thin films made from a fine mesh of conductive nanotubes. These thin films could also be useful in some displays, replacing indium tin oxide electrodes, which are becoming more expensive as demand for them grows. Because of the superior conductivity of metallic carbon nanotubes, these films have the potential to improve the performance of organic solar cells, which also require transparent electrodes, Martel says.

The ultimate goal of computer chips that use single-nanotube transistors is years away--researchers still face the challenge of developing techniques for assembling millions of densely packed transistors and the connecting wires in intricate circuits. But the Northwestern advance could remove one of the basic obstacles to developing such nanotube-based electronics. "We find this very exciting," says Martel, "because we can push forward with our work."