Graphene’s New Rival
Molybdenum disulfide helps graphene transistors work better—and it makes good nanocircuits on its own, too
Photo: Andras Kis
OOH, SHINY! Slices of molybdenum disulfide [above] can be easily pulled off the bulk material and then fashioned into nanocircuits.
Graphene has become the darling of the postsilicon crowd in the eight years since Andre Geim and Konstantin Novoselov isolated it by ripping Scotch tape off a chunk of graphite. But there are other two-dimensional nanowonders. Molybdenum disulfide (MoS2), which can be pulled off a block of molybdenite through the same process, could offer new approaches to making high-speed logic circuits—on its own or in combination with graphene.
“We spent the last seven, eight years looking at how to make transistors out of graphene,” Geim says. “But there was an elephant in the room—that you can’t really switch off current in graphene.”
Graphene’s single atomic layer of carbon atoms allows electricity to flow rapidly, promising circuits that work far faster than silicon transistors. But it lacks a bandgap, meaning that it’s hard to shut off the flow, making the on and off states of digital logic nearly impossible. Several approaches—using nanoribbons, quantum dots, or double layers of graphene—have been tried, but they are difficult, poorly developed, and tend to undermine the speed advantage.
So Geim and his colleagues at the University of Manchester, in England, tried a different design. They built a field-effect transistor with a vertical heterostructure—two layers of graphene separated by MoS2 or boron nitride. Like graphene, those materials are considered two-dimensional because they’re as thin as they can be; MoS2, for instance, is a single atomic layer of molybdenum sandwiched between two single layers of sulfur. The material acts as a barrier, preventing charge from crossing. So ordinarily, the transistor is in the off state— no charge flows from one graphene layer to the other.
For the on state, the researchers raise the voltage of one of the graphene layers. That boosts the energy of the electrons in that layer, causing them to tunnel through the barrier into the other graphene layer. Since tunneling is by nature very rapid, the process should make for very fast circuits, Geim says. The device area could be scaled down to as small as the latest lithographic techniques allow, which is better than today’s silicon circuits, Geim claims.
Using boron nitride, the ratio between on and off was only about 50, but with MoS2 it was about 10 000, sufficient for some logic circuits. But Geim isn’t wedded to MoS2. The experiments, published in Science in February, were only a proof of concept, and there may be some material that works even better.
Image: Andras Kis
ALTERNATIVE IC: Molybdenum disulfide transistors have been formed into logic gates.
But MoS2 might work just fine on its own, according to Andras Kis, an assistant professor in the Laboratory of Nanoscale Electronics and Structures at École Polytechnique Fédérale de Lausanne, in Switzerland, who presented research on MoS2 devices at the American Physical Society’s March meeting in Boston. Instead of using MoS2 as the insulating filling of a three-layer sandwich, his team used a single layer of the material as the semiconductor channel that connects source and drain in a transistor, achieving an on-off ratio of 100 million. Combining two to six of these transistors, he made integrated circuits—inverters, NOR gates, and other logic. He says it should be possible to make channels smaller than 10 nanometers in length, perhaps as tiny as 5 or 6 nm, smaller than silicon is likely to ever get. The smaller transistors are, the more can be crammed onto a chip.
Unlike graphene, MoS2 does have a bandgap, which makes switching more straightforward. In fact, at 1.8 electron volts, its bandgap is higher than silicon’s (1.1 eV). It’s actually higher than you can get in graphene, if you make the graphene into thin ribbons, as some researchers do, Kis says. The higher bandgap means switching would require less power, which could be attractive for mobile applications.
“It’s probably not going to be as fast as graphene, because it doesn’t conduct that well, but I think it will be comparable to silicon,” Kis says. And if the circuits are smaller with the same speed and lower power consumption, that’s still an improvement over silicon.
Researchers have shown that MoS2 also resembles graphene in having attractive mechanical properties, though not quite to graphene’s extremes. It’s 30 times as strong as stainless steel, where graphene is 100 times as strong. It’s also very flexible; it can bend to a deformation of 10 percent without breaking, whereas most semiconductors (graphene excluded) break at less than 1 percent. “It’s more like a polymer in that respect,” Kis says, adding that it might be used in flexible electronics. He’s not expecting MoS2 to displace graphene as the new wonder material. But it still may help pave the way to a postsilicon future for electronics.