New technology challenges basics of computers

Challenging a basic tenet of the semiconductor industry, researchers at Hewlett-Packard have demonstrated a technology that could replace the transistor as the fundamental building block of all computers.

The devices, called crossbar latches, could be made so small that thousands of them could fit across the diameter of a human hair, enabling the high-tech industry to continue to build ever-smaller computers that are less expensive than their predecessors.

For years, engineers have been able to pack more and more smaller transistors onto a fingernail-size silicon chip. The rate of integration, first predicted by Intel co-founder Gordon Moore in 1965, has driven computer performance and prices for more than 30 years.

But the pace of Moore’s Law can’t continue forever, and the high-tech industry has been scrambling to develop workarounds for the day – expected in a decade or so – when transistor dimensions become too small for the materials commonly used today.

The smallest features of today’s silicon-based transistors are about 90 nanometers long, a nanometer being roughly one hundred-thousandth the width of a human hair. The crossbar latch, by comparison, can work in a space of about 2 to 3 nanometers.

The HP research, reported in the Journal of Applied Physics, scraps the transistor entirely. In its place is basically a series of platinum wires crossed opposite directions. At the junctions are molecules that in the HP research happen to be steric acid.

Like in a transistor, an electrical signal that passes through a crossbar latch is manipulated to perform logic functions. The latest research shows that the technology also can be used for amplifying a signal, allowing multiple functions to be applied.

“The power of this device is not when it’s by itself. It’s when it glues together other pieces of logic,” said Duncan Stewart, another HP Labs scientist and study co-author. “As soon as you’re able to do that, we call that a computer.”

The researchers have not glued together multiple crossbar latches, though they say it’s something they’re continuing to pursue. They expect it to be commercially viable as early as 2012. The latches are formed through a specialised stamping process for nano-sized imprints.

They also must persuade an industry built on transistors that an alternative technology can be just as effective, said Stan Williams, director of Quantum Science Research at HP Labs and another of the paper’s co-authors.