Researchers at IBM had demonstrated an approach similar to Dai's in 2001 in a Science paper. To get rid of metallic nanotubes, they passed a high current through the devices, blowing out the tubes like a fuse. But the plasma etching method has an advantage. With current, one has to treat each device individually. But with heat, one can treat an entire wafer, which may contain several devices, at once. "With the [IBM] method, you have to go and individually address every single device," says Mark Hersam, materials-science and engineering professor at Northwestern University. "I can imagine some ways of doing that efficiently, but there's no doubt that you can do this Dai approach efficiently."

Hersam recently reported an advance (see "Nanotube Computing Breakthrough") in sorting nanotubes according to their electronic properties and diameters, yielding 99-percent-pure sources of semiconducting nanotubes or metallic nanotubes.

It's not easy to take presorted nanotubes and assemble them into devices, says Richard Martel, a chemistry professor at the University of Montreal. "This [Dai] method doesn't get there, but it's much closer to the reality," he says. "It's easier for translating to large-scale applications."

The technique's main constraint is that it works in a narrow diameter range, Hersam adds. As the Stanford researchers show in their paper, the plasma selectively etches metallic nanotubes only when the molecules are between 1.4 and 2 nanometers wide; the method gets rid of all nanotubes narrower than 1.4 nanometers, and leaves intact metallic and semiconducting nanotubes that are wider than 2 nanometers. Dai acknowledges this limitation and notes that "the method will have the most potential if the starting material is carefully chosen."

Martel cautions that we will need to see other breakthroughs similar to Dai's before nanotube electronics becomes a reality. But that future might not be too far off, Dai says, because other research groups have developed various elements of the nanotube-chip puzzle. "I think now it looks like we have answers for all of them," he says. "It's a matter of putting them together."