IBM researchers have created transistors out of carbon nanotubes that can outperform similar silicon transistors, a development that helps build the case that carbon may one day become a building block of computing.
In an article to be published on Monday in the journal Applied Physics Letters, IBM researchers outline how transistors made of carbon nanotubes--long, thin strands of carbon molecules--delivered more than twice the amount of electrical current at a faster rate than cutting-edge transistors made from silicon and metal, the basis for chips today.
Increased current leads, potentially, to faster transistors and integrated circuits. And since transistors and integrated circuits are the building blocks of chips, the results imply that carbon could someday become the foundation for tomorrow's computers.
"They outperformed silicon transistors," said Phaedon Avouris, manager of nanoscale science at IBM. "There are a number of (performance) improvements."
Although nanotechnology--essentially, the science of building things on a molecular level--is in its infancy, it could over the ensuing decades become crucial to a wide array of industries, say advocates and many researchers. General Motors, for instance, is trying to develop stronger materials for cars using methods from nanotechnology.
In the field of high tech, IBM, Hewlett-Packard and other companies are experimenting with these techniques to build microscopic circuits. Eventually, these circuits could be used to create inexpensive sensors for detecting gas leaks or other environmental hazards, or, some two decades down the line, small data storage devices or computer chips.
Why all the hype about carbon nanotubes? Computer processors are becoming so dense with circuits that many believe traditional methods of designing and making chips will become economically untenable in 15 to 20 years.
In the case of nanotubes, molecules arrange themselves into patterns like snowflakes, so individual chip circuits would no longer have to be drawn--a drastic change that would dramatically drop the labor, factory and equipment costs in the semiconductor industry.
Performance would also improve because of the way electrons travel across the tubes, said Avouris. In traditional wire circuitry, designers have to deal with resistance--that is, forces and obstacles that slow down electrons--so electrons get scattered, which creates heat and loses energy.
By contrast, nanotubes are extremely thin and long--so thin, in fact, that electrons can't be deflected sideways, and can be stopped or reversed only with great difficulty.
"The collision has to be very strong to reverse the direction of the election. In principal, there are not many things that are strong enough to do that," Avouris said. "You only have two directions of propagation."
Theoretically, nanotubes would allow engineers to build chips that would require less electricity (because less is lost in transmission) or could deliver more performance for an equal amount of energy.
Of course, the benefits are still hypothetical. Reseachers, for instance, recently determined that there is an energy barrier of sorts at the thin, contact end of the tubes that doesn't occur with standard silicon transistors. The ability of molecules to arrange themselves has also not fully been plumbed.
"The basic science is still not totally understood," Avouris said. "Nature does use self-assembly, but nature had a research and development time of over 2 billion years."
Creating carbon nanotubes remains a long, difficult process. Carbon Nanotechnologies, a start-up out of Rice University, grows the tubes in a chemical bath by placing a metal particle inside a carbon-heavy petrochemical bath. After everything gets heated to more than 1,000 degrees centigrade, freed carbon molecules begin to form a strand off the metal.
IBM acquires the strands from Carbon Nanotechnologies and inserts them into its own chips.