|September 21, 2000|
Carbon bits to revolutionize computer construction
By William J. Cromie
A new way of building computers involves the world's strongest material in the form of exotic tubes 100,000 times thinner than a human hair. Called nanotubes, they are a hundred times stronger than steel, able to bend without breaking, and efficient at conducting electricity. But to see them you have to look into a powerful microscope. "Nano" means billionths of a meter, or a hundred-millionths of an inch -- smaller than the usual meaning of small.
In the Chemistry Department at Harvard University, for example, you see two thin lines under the microscope, one nanotube crossing another at right angles.
A wisp of electric current bends the top tube down to meet the bottom one, forming a switch about a hundred times smaller than those in the fastest existing computers. Arrays of such switches can, in principle, be connected to form a computer memory, or logic circuits that enable the computer to make choices.Such nanomachines could combine high computing power with extremely low electrical requirements. They might make possible a new realm of portable, wearable, or even implantable computers that aid in the detection and monitoring of diseases, says Charles Lieber, Mark Hyman Jr. Professor of Chemistry. They also might make it possible to carry all the information now in your office files on a gadget no bigger than a wristwatch.
Lieber and his research team of Tom Rueckes, Kevin Kim, Ernesto Joselevich, Greg Tseng, and Barry Chueng have also pioneered the fabrication of nanotubes for a variety of other possible applications from probes that detect single molecules to construction materials for turbine engines and body armor.
The tubes are one response to a cloud of concern building on the computer horizon. In 1965, Gordon Moore, one of the founders of Intel, now the world's largest maker of computer chips, predicted that the number of electronic switches and other devices put on a single chip would double every 18 months. He didn't say for how long, but his prediction still holds. In 1964, a chip about 1-inch-square held 32 devices connected by wires. The Pentium III chip, introduced last year by Intel, holds 28 million circuit elements. Pentium IV, expected to reach the market later this year, will boast 42 million elements.
That kind of growth has no parallel in industrial or economic history, but it can't last. To add more devices on a chip not much bigger than a thumbnail, their sizes must be reduced to nanometer dimensions. "At these sizes, devices presently in use reach fundamental physical limitations that prevent them from functioning reliably," Lieber points out.
Also, as numbers of devices double, so does the cost of making computer chips. Production lines for chips like the Pentium III now cost several billion dollars. Many experts are convinced that the end of doubling growth will come around 2010, perhaps earlier.
Memories from sootExisting chips and the devices they hold are made principally of silicon, the main ingredient of sand and one of the most abundant elements on Earth. The search for a replacement centers on materials of nanometer size with comparable, or superior, strength and electronic properties. Many molecules, the smallest particles of any complex substance like a metal or a protein, are of this size. In theory, computer chips an inch square could hold billions, even trillions, of such molecular-scale switches. Today's Pentium-style chips could be reduced to the size of a needle tip.
Much effort has gone into trying to construct electronic devices out of organic molecules (which make up fuel oils and living things). But Lieber thinks organic molecules may be too unstable to form a reliable computer memory at the level of a single molecule. Research teams at several university and industrial labs are, however, trying to fabricate electronic switches and other devices based on large numbers of organic molecules.
Nanotubes can be produced from vaporizing graphite at high temperatures. The results looks like soot, but under a powerful electronic microscope you can see tiny carbon tubes that resemble tightly rolled chicken wire, but are only 10-20 atoms in diameter and thousandths of an inch long. It's hard to believe anything like that has the strength, elasticity, and electronic properties to revolutionize computer construction.
With the help of colleagues in the Department of Chemistry and Chemical Biology and the Division of Engineering and Applied Sciences at Harvard, Lieber managed to make both switches and wire connections from nanotubes. As described above, two nanotubes crossing at right angles make a simple and effective switch. So far, the team has produced only single junctions, but Lieber feels these junctions can be assembled into workable computer memory and logic devices before silicon devices hit their limits.
Hybrid approachesNanotube switches could -- theoretically -- operate at startling speeds, the individual junctions opening and closing billions of times a second or faster than their fastest silicon competitors. Lieber is aiming for 100-200 gigahertz, or 100-200 billion cycles per second. (According to Intel, its Pentium IV chip will operate at 1.4 billion cycles per second.)
As another advantage, nanochips might be produced in facilities with a certain amount of dust and dirt. Fabrication of existing microchips is done in special "clean rooms," where workers wear dust-free clothing and pressure keeps air moving from inside to outside of the facility. "Dirty" nanochips would have more defects than those in today's machines, but operating programs might be written to keep that from being a serious problem.
Of course, there are other paths to molecular computers being followed by other universities, and companies like IBM and Hewlett-Packard. Lieber is considering devices that would marry silicon nanowires with carbon nanotubes. Such a hybrid could make it easier to connect together switches on the same chip and chips in the same computer.
The military has expressed deep interest in nanotubes. The Defense Advanced Research Projects Agency (DARPA) recently solicited proposals for research in this area. Although he had been working with nanotubes for several years, Lieber did not respond. But DARPA contacted him and asked him to come up with something.
"I did, and DARPA support has led to our making the nanotube junction switches," he says. "The approach we used opened our eyes to more areas where nanotubes might be useful. And we also had a lot of fun."
Copyright 2000 President and Fellows of Harvard College