March 2002

The Nanotube Computer

Continued from page 2

By David Rotman

Samsung expects to have full-color prototypes capable of the resolution needed for high-definition television this winter and an 81-centimeter TV ready for the market by late 2003 or early 2004, says Jong Min Kim, vice president of research at the Samsung Advanced Institute of Technology. Key to success in the $100 billion display market, he says, will be getting manufacturing costs of the nanotube TVs low enough that they can compete with cathode-ray-tube models. "First we will try to attack the TV market, then we'll go after the computers," says Kim.

Instant Turn-on

Far from the large corporate labs working on nanotube TVs is a tiny Woburn, MA-based startup called Nantero, whose employees hope to take on another multibillion-dollar market-computer memory. Sitting in Nantero's conference room, which also serves as a front entrance, lobby and kitchenette, cofounder and chief scientist Tom Rueckes seems both anxious and excited. And well he should be. The year-old company is promising a high-density nanotube-based memory that would revolutionize the market. And it claims it will have this breakthrough working within two years. "Imagine," says Rueckes, "having several gigabits of memory at your fingertips that is instantly on."

Indeed, the most attractive aspect of the Nantero memory is that it will be "nonvolatile." Conventional dynamic random-access memory (DRAM), the short-term electronic memory that a computer uses to run its operating systems and programs, holds information only as long as the power is on. That's why a PC needs to be booted up: the machine has to rewrite stored information from the hard drive onto the electronic memory. Nonvolatile memory means never booting up again. Eventually, if the storage capacity of nonvolatile memory chips gets large enough, they could make magnetic hard drives obsolete.The best existing DRAM can hold about one gigabyte of data. Within two years, Nantero expects to have a nanotube-based nonvolatile memory chip with several gigabytes of capacity.

The nanotube memory is based on an ingenious, though strikingly simple, design that Rueckes came up with while a PhD student under Lieber at Harvard. An array of parallel nanotubes is suspended just a few nanometers above a perpendicular array lying on a substrate; each intersection of the cross-arrays represents a potential bit of memory. When an applied electrical force stretches a tube in the top array close enough to a lower tube, they physically bind and a current can flow between them; the switch is on and stays on even when the power is turned off. Because each bit of memory is so small, a centimeter-sized chip based on the design could have, in theory, terabits (a trillion bits) of nonvolatile memory.

The goal is to turn this laboratory design into real technology as quickly as possible. Rueckes declines to detail exactly what has been built so far, except to say that "components of it are working." But he adds that the strategy is to integrate nanotube memory with conventional electronics. "We want to come up with a product that can be manufactured with existing technology," he says.

Such nonvolatile memory would change how people use their computers, doing away with those tedious minutes spent booting up. But the real prize in nanoelectronics-the one that will make people truly forget about silicon-is the logic circuits that are the brains of computers. Moore's Law, the oft-cited 1965 prediction by Intel cofounder Gordon Moore that the number of transistors on a chip would double every 18 months, has held for more than three and a half decades. But experts predict that within a decade or so, it may well be impossible to make silicon transistors small enough to continue to uphold Moore's Law.