It's no secret to chipmakers that a big change is coming. Georgia Tech physics professor Walter de Heer describes it this way: "Moore's Law is coming to an end in a decade or so. Chip technology won't grow every 18 months as it has for 20 years. You won't see constant processor improvement within the next 10 years. . . . Silicon can't keep up."
Of course, the chip industry isn't doomed. All it needs is a new material, something smaller than silicon that can still handle all the hard work it takes to be a modern transistor.
For years, carbon nanotubes were heralded as the solution. Nanotubes are made from rolled up sheets of carbon, only a single atom thick. The problem is, nanotubes are difficult to turn into chips. "After ten years working with nanotubes, no one has figured it out," says de Heer, who worked with nanotubes in the 1990s.
De Heer took another look at the flat carbon sheet—called graphene—and calculated that instead of rolling it into tubes, he could cut it into narrow strips with the same properties. He estimates that graphene can be 100 times faster than silicon and may one day hit terahertz range. It may even be suitable for quantum electronics.
Making graphene is more complicated than simply scribbling with a pencil lead on paper. A silicon carbide crystal is heated in a vacuum furnace until the surface is converted to epitaxial graphene. A molecule-thick layer is peeled off with tape. Then the fun begins.
With graphene strips, "the problem of connecting nanotubes and putting them where we want is eliminated," says de Heer. "You can tailor any shape." The secret is in the cutting. You can't just take scissors to a two-dimensional sheet of graphene. It's done with microelectronic lithography techniques, like those already used in current silicon fabrication labs (fabs). His team at the school's own microelectronic fab has developed a way to do just that.
Is de Heer bullish on graphene? He is when he says it "overcomes a lot of show-stopping problems found with silicon. More current can go through it, it doesn't heat up as much, and you can go down to nanometers or smaller—and it still works." Work continues in other labs on carbon nanotubes and other potential silicon replacements. "Every year, there's a new scheme," says de Heer, but he says graphene is the horse to bet on.
At the same time, his enthusiasm is tempered by a desire to avoid hype. Graphene isn't an instant panacea. "There's no question in my mind that it'll take ten years to develop these technologies, to get the lithography to the point of making devices that are comparable to silicon devices," he says. "It'll be really hard."