It took a while, but the explosion of research that followed IBM's 1986 discovery of high-temperature superconductors is on the verge of paying off. On Aug. 10, Superconductor Technologies Inc., a Santa Barbara (Calif.) startup founded in 1987 to harness the new technology, is set to unveil a prototype of the world's first superconducting "package" for computer chips. "It's clearly a good first start in exploiting high-temperature superconductors in electronics," says Richard E. Harris, a group leader at the National Institute of Standards & Technology's superconductor research laboratory in Boulder, Colo.
High temperature, in this case, is relative. The material that STI uses for its chip package--a thallium-based compound developed at the University ofArkansas--becomes superconducting around 270F. Still, to physicists, that's a lot warmer than the 450F normally required for superconductivity.
THIMBLE THERMOS. That warming trend is big news to the computer industry, since it brings computer makers a major step closer to producing business machines that are 10 to 100 times faster than what's available today. Indeed, those speedy new machines could be commonplace commercially within a decade, some experts predict.
To the military, such cryogenic electronic systems--those that are cooled to 321F--have already become old hat. Almost every major weapons system, from Bradley fighting vehicles and M-1 tanks to F-14 fighters, carries a "cryo-electronics" component, sometimes produced by STI. "There are plenty of systems out there today, maybe 300,000, that operate at cryogenic temperatures," says James G. Bybokas, STI's vice-president for product marketing.
The technology faced harsh field tests in the gulf war. Many of the military's high-tech weapons, including night-vision systems, work only when cooled to cryogenic temperatures by tiny high-tech refrigerators--some no bigger than a flashlight.
Now, the challenge is to make the supercold, superfast technology cost-effective enough to use in commercial computers. That's beginning to happen, thanks to pioneering work in the technology by the Defense Advanced Research Projects Agency. DARPA is pouring money into superconductivity research, hoping to boost the speed of military electronics by a factor of 1,000. Advances of that magnitude may be possible only by putting chips in superconducting packages.
TIGHT PACKAGE. That's where Superconductor Technologies' development comes in. The company's achievement revolves around a new way of packing multiple chips into a single, tiny package called a multichip module (MCM). Until now, advances in computing speed have stemmed mainly from etching thinner circuit lines on silicon chips. Smaller lines mean the maze of circuitry can be scrunched closer together, so electronic signals have shorter distances to travel on each chip.
Soon, though, just shrinking the circuit lines on individual chips won't do the trick. The signals may take only a few billionths of a second to shoot across the gap between chips mounted on circuit boards, but that's too slow for tomorrow's superfast desktops. MCMs get around this bottleneck by butting chips together inside the same package. Big computers have used this technique for years--with telling results. Digital Equipment Corp. says its VAX 9000 mainframe runs twice as fast with MCMs, each housing as many as 76 chips, than it would without them.
STI has a powerhouse team of scientists working to make the technology both common and inexpensive. Many of them are researchers at the University of California at Santa Barbara--including J. Robert Schrieffer, a physicist who shared the 1972 Nobel Prize in physics for his 1947 work in superconductivity.
To fund its research, STI last November got a share of the latest DARPA morsel for the field: a $10 million, three-year program to develop superconducting MCMs. Headed by E-Systems Inc. in Dallas, the effort involves a consortium of universities, national laboratories, and companies. MCMs may not be the stuff of everyday discourse, but they clearly enjoy broad support on Capitol Hill. Congress virtually doubled the project's funding in January, to $19.5 million.
Now, since STI produced a working prototype so rapidly, DARPA hopes to plow still more money into the program. Senator Dale L. Bumpers (D-Ark.) says he will support an increase in funds. MCMs, he says, could "exert a profound influence" on the future competitiveness of the U.S. With the electronics industry battling for U.S. markets that some estimate will reach $1.3 trillion by 2000, superconducting at 321F sounds positively sunny.