Supercooled Computers

The essential characteristic of all modern computers is speed. Their prodigious memories can be probed with split-second precision; they can race through reams of complex equations with astonishing agility. Their swift skill is made possible by a battery of relatively simple devices, transistors that can turn an electric current off and on in as little as a billionth of a second.

In effect, those switches speak the "yes-no" binary language of computer technology. Their simple answers can be combined to solve intricate problems. But fast as such combinations can be made, computer speed is often not fast enough. The big machines strain to their limit to handle the demands of space travel; they are also too slow to process in time the vast amount of meteorological data necessary to make the detailed and accurate five-day weather forecasts the U.S. Weather Service would like to achieve.

Now help may be at hand. After five years of effort, IBM's research labs have developed an electronic switching device that can be turned on and off in less than ten trillionths of a second --more than 100 times faster than the fastest transistor used in computers. What is more, IBM's development requires only about one ten-thousandth of the power necessary to run these transistors; it gives off only a tiny fraction of the heat they radiate. And it is transistor heat as much as switching time that limits a computer's skills. For when transistors are packed closer together in order to speed up the flow of signals between them, the risk of overheating is sharply increased.

IBM's switch is based on a phenomenon first predicted in 1962 by a British scientist named Brian Josephson, who was only 22 at the time. While studying superconductivity,* the Cambridge graduate student determined mathematically that pairs of electrons would "tunnel" through material that is normally an electrical insulator if it is thin enough and sandwiched between two superconductors. If the flow of electrons through the insulator were kept below a certain critical value, he found, there would be no difference in voltage from one side of the insulator to the other. (At normal temperatures, an electric current never flows unless there is a voltage differential.) Josephson also predicted that if an external magnetic field were applied to the junction, a voltage drop would appear.

Later verified by experiment, the so-called Josephson effect has been widely used to construct extremely sensitive laboratory measuring devices, including a magnetometer that can detect fluctuations in a magnetic field only one five-billionth as strong as the earth's. But IBM scientists found a more practical use. They knew that they could produce a voltage drop across a Josephson junction by applying a weak magnetic field; generating that field would require only a fraction of the energy required to switch a transistor. Furthermore, the presence or absence of that voltage across a Josephson junction could be used to represent the same "yes" or "no" information conveyed by a transistor.

For competitive reasons, IBM will not reveal the precise chemistry of the lead alloys used in its junctions. In fact, the company is cautiously refraining from predicting when they will be used in practical computers; many design problems must be overcome before computers can be built to operate at superconducting temperatures. Nonetheless, IBM's laboratory triumph and continuing research by the world's largest computer manufacturer suggests that high-speed, supercooled electronic brains are not far in the future.

* The disappearance of electrical resistance in certain materials when they are cooled to within a few degrees of absolute zero ( --459.7DEG F.).

This file is automatically generated by a robot program, so reader's discretion is required.