Theorists calculate that nanotube transistors will have more functionality at reduced size

Francois Leonard, a theoretical physicist in Thin-Film and Interface Science Dept. 8721, and his IBM collaborator have discovered one more reason to further develop nanotube transistors.

Over a two-year effort to understand and model these new-material laboratory curiosities, they found these transistors work differently from conventional devices and, in fact, offer an additional way to switch the transistor on and off.

The promise of more functionality at a reduced size, Francois says, "is a further motivation to try to make nanotube transistors even smaller."

Their findings, published in Physical Review Letters, also offer a new approach to modeling that will be of interest for predicting performance of nanoscale devices in general.

Conventional devices can be switched off by raising voltage at a gate between two electrodes. The research team shortened the conventional distance between electrodes some 100 times, to just 10 nanometers. They modeled a device in which the electrodes were linked by a strong, thin filament of graphite-like carbon, rolled into a nanotube no more than two nanometers in diameter.

The new model revealed that increasing voltage at the gate first turns off the transistor as in a conventional device, then switches it back on again. This occurs because some electrons tunnel through a quantum state and move across the gap via gated resonant tunneling — which becomes the added functionality possible at this smaller scale. Increasing the voltage even more then creates a negative differential resistance, turning the circuit off again.

"The reduced dimension actually gives you additional functionality," Francois says, comparing the phenomenon to a quantum dot. "By controlling the gate voltage, we can make electrons pass through just one (quantum) level." Furthermore, unlike silicon, devices made with nanotubes would not need to be "doped" with impurities. The work involved new ideas in how to calculate current in a device in which electrons move in single file. "How do you calculate current when devices are governed by single electrons?" Francois asked.

Existing modeling tools were not adapted to the problem because they were more statistical in nature, essentially allowing behavior to be predicted by averaging the flow of many electrons.

Francois’ co-author is Jerry Tersoff of IBM’s T.J. Watson Research Center, a fellow theoretical physicist. The calculations, which took a couple of years to develop, were presented last spring at the San Francisco meeting of the Materials Research Society and are the subject of an invited talk at the International Conference on Computational Nanoscience and Nanotechnology this month.

Carbon nanotubes were first discovered in the laboratory about 10 years ago and are novel materials for several reasons. They are very strong and can be made in long filaments. Depending on their atomic structure the material can behave either electronically like a metal or a semiconducting material. The first nano- tube transistor was demonstrated in the lab more than three years ago. However, creating just one transistor takes some time, so potentially packing more circuits into small spaces with these materials is still very exploratory.

On the other hand, Francois says, since nanotube transistors have been demonstrated, using them in practical devices "is not that far-fetched — it is a matter of time and effort."