Creative Destruction

by Dan Murray

Published May 02, 2001



IBM researchers announced last week that they have made the smallest transistors from tiny carbon tubules. These chemically constructed tubes are only one nanometer in diameter, or 200 times thinner than the finest lines in existence on chips today. If all goes well, commercialization could begin within three years, insuring the progression of computers becoming smaller and faster.

As computer electronic components and their interconnecting metal pathways shrink toward the molecular level, the shorter the distance that electrons need to travel, thus the faster they go. The early work begun at IBM in Holland two years ago was systematic trial and error experimentation. Chemist Phaedon Avouris said, “It is an important first step in learning how the characteristics of carbon might be better than silicon.”

Carbon crystal formations can be either a conductor of electricity, like a metal wire, or a semiconductor necessary for making transistors. It is stronger at smaller sizes and electronically versatile. Some adaptations could be in high heat environments such as sensors on a space vehicle’s reentry heat shield.

“Silicon is already pretty fast. We’ll need an advantage factor of 10 [times improvement] or it’s not worth it,” says Paul McEuen, physics professor at Cornell University. Single-molecule fiber carbon nanotubes, although incredibly small, can be stretched and still have the tensile strength 100 times greater than an equal weight of steel.

IBM researchers chemically altered the characteristics of carbon as a conductor into the material of transistors by a process IBM calls “Creative Destruction.” They first apply a small voltage to the crystalline wafer which switched off the semiconductors. The remaining conductive carbon tubes are vaporized by a powerful jolt of current sent through the remaining metallic tubes.

The result is the creation of field effect transistors, the most commonly used in today’s electronics. They are almost 500 times narrower than has been used before.

It’s just a beginning demonstration. There is no process yet devised to replace lithography with a chemical process to impress this small size circuitry upon a chip of carbon. A new computer’s architecture based upon carbon will evolve with it’s own set of performance characteristics.

Speed of the silicon chip is impaired not within the transistor but along the bumpy metal lines connecting them. The use of carbon nanotubes as conductors on the same material has not yet been devised. “It’s hard to imagine using the technique of blowing up the wrong tubes,” says Professor McEuen.

The theorists believe that the microstructure should allow electrons to flow through the tubes “like cannonballs” unimpeded by the imperfect surface defects of metal over processed silicon.

Scientists are now measuring the energy levels of an electron, called the “bandgap.” Silicon’s bandgap is one half an electron volt while carbon nanotubes can be adjustable, designed for differing applications, from a full volt to a very small fraction of one volt. The carbon advantage is showing promise.

The researchers nested multiwalled tubes, one inside the next, for low frequency signals. For high-frequency radiation uses, they blew off the outer shells to make a tube small enough to have a big enough bandgap.

Thomas Theis, director of physical sciences for IBM research, says, “For maybe a couple of years, we’ll be examining how small these can be made and how well they can function. Then the focus will shift to control the chemical synthesis of nanotubes much better. We could envision building a key part of a device with chemical synthesis instead of lithographic patterning.”

IBM’s Zurich laboratory, earlier this month, successfully assembled metallic carbon nanotubes in orderly crystals. The next step is an entire logic circuit of carbon made with molecular-sized building blocks miniaturized in one piece.

“Every device works on a certain principle, one that does not necessarily apply on a smaller scale,” says Avouris. “A truly atomic-scale device has to come with its own operating principle, and we don’t know of any now.”