Sematech Makes a Device 1/200th the Size of a Hair

IMAGINE driving cross-country in one minute in a car that cost less than a dollar.

That's what would be possible if cars had improved as much as computers have in the last 15 years, says William George, chief operating officer of Sematech, the Austin-based government-industry semiconductor consortium.

Now, Sematech says, it has achieved a breakthrough that puts the United States back in the driver's seat of the global computer industry. This success, Sematech executives say, validates their company as a model government-industry partnership.

The executives caution, however, that the Sematech model should only be followed when an industry is vital to the US economy or national defense. "No buggy-whip industries," remarks spokesman Miller Bonner.

The electronics industry meets both criteria, Sematech executives say, being the largest employer in the nation and a maker of components for the advanced weapons systems used in the Gulf war.

Last Thursday the consortium announced that it had demonstrated the ability to manufacture integrated circuits with electronic device widths of 0.35 microns. And it did so using only American-made tools.

To reach this achievement, Sematech has spent $1 billion since its inception five years ago. Half was invested by a handful of private corporations that represent 75 percent of semiconductor manufacturing in the US.

The other half was paid by the federal government through the Defense Advanced Research Projects Agency, which has a $1.5 billion budget to sponsor research by universities and national laboratories.

DARPA was founded in the 1950s to prevent future occurences of "technological surprise" like the launching of Sputnik by the Soviet Union. Today its mission is "to maintain technical superiority over potential adversaries," a DARPA spokeswoman says.

US companies have recently regained the lead from Japan in both computer chip sales and in sales of the machines that make the chips. Japanese companies had become No. 1 in chip sales in 1985. But last year US companies notched sales of more than $25 billion for a market share of 43.8 percent, compared with 43.1 percent for Japanese manufacturers, VLSI Research Inc. estimates.

US toolmakers saw their share of chipmaking equipment sales recover last year to an estimated 53.4 percent of the $10 billion market from a low of 43.9 percent in 1990, according to VLSI Research.

Sematech executives say the consortium deserves at least some of the credit for that because of the research it has sponsored and the communication and cooperation it fostered among its member companies.

"They have definitely augmented our competitiveness, not only domestically but globally," says Karen McLennan, a spokeswoman for Lam Research Corporation in Fremont, Calif. Lam is headed toward $200 million in sales of its chipmaking tools, up from $170 million last year, she adds.

When Sematech was founded it set a goal of demonstrating by this year the capability to make semiconductors with device widths of only 0.35 microns, or 1/200th of a human hair. If a thumbnail-sized integrated circuit were expanded to the size of Washington, D.C., a device with a width of 0.35 microns expanded proportionately would be the size of an apple.

Leading-edge companies currently make chips with 0.8 micron device widths but are preparing to step down to 0.5 microns wide. The 0.35 micron width will be the standard for the subsequent generation of chips in the mid-1990s. And American chipmakers will have the capability first, using US-made tools, Sematech executives say.

The smaller size will bring manufacturing costs down while making possible the commercial production of 64-megabit and 256-megabit memory chips far more powerful than are available today. Such chips will be the basis for new products, such as a pocket-sized personal communicator that works anywhere on the planet, electronics writers have predicted.

Next, Sematech plans to reduce device widths to 0.25 microns by 1994, and to 0.18 microns beyond that. "The technology hurdles are immense," Mr. Bonner says. "We're talking about circuitry measured by the number of atoms wide."

Two hundred scientists from industry, academia, and government came together in Dallas last month to discuss what goals the computer industry should work toward through 2015. Now being reviewed by other scientists, the program will be announced next month. Sematech executives helped define that program and expect the consortium to play a role in carrying it out.