Helping microchips get smarter

RTP companies benefit from $1 billion an industry group has funneled to research

Twenty-four years ago, the U.S. semiconductor industry was at a crossroads.

Companies such as Hewlett-Packard, Intel and IBM were losing ground to competitors in Japan that were producing higher-quality computer chips. Executives were alarmed to see that spending on basic research was declining in the United States.

So industry leaders teamed up to create Semiconductor Research Corp.

Since 1982, the consortium has quietly funneled more than $1 billion in government and private funds into research in electronic-chip design, manufacturing and testing at universities around the world.

SRC bills itself as the best-kept secret in the chip industry. Its members include Advanced Micro Devices, Novellus Systems, Texas Instruments, IBM and Intel, and its research has driven the evolution of the newest chips that power cell phones, coffee makers, computers and cars.

"We have become recognized as the pioneers of collaborative research," said Larry Sumney, SRC's chief executive since its inception.

It's also a well-kept secret in the Triangle.

After the Semiconductor Industry Association's board created SRC in 1982, they started hunting for a headquarters location independent of the heavy hitters in the industry. With three major universities, a strong electronics advocate in Gov. Jim Hunt and the newly formed Microelectronics Center of North Carolina, the Triangle -- Durham, specifically -- became the SRC's home.

Today, SRC's mission is just as critical as it was two decades ago, said Gary Smith, chief analyst for design and engineering at Gartner Dataquest. After a period of major change in the early 1980s, basic research on semiconductors and microelectronics dwindled in the 1990s.

"I call it the cream-puff era of semiconductors," he said. "Everything was incremental, predictable. You had a road map everyone followed."

Then engineers shrank the transistors on chips to a minuscule 130 nanometers -- about one two-thousandth of the width of a human hair -- around the turn of the century and "we started running into serious product and process engineering problems," Smith said.

Many basic research labs had closed in the research-light previous decade, and only a few companies still had the research facilities to address those kinds of issues, he said. So the industry turned to SRC.

The organization listens to the industry's needs, identifies and funds university projects to meet those needs, then communicates the results to its members as quickly as possible.

"We're doing better now than we've ever done," Sumney said. "There's a continual reason for us to exist, because no single company can fund a research program that covers the full range of research that is needed."

SRC gives its members leverage: By paying an annual membership fee that ranges from $100,000 to $7.5 million, companies and government bodies can take advantage of the results generated by their pooled resources. Some of SRC's research dollars recently have been used to develop chips that work despite manufacturing imperfections or component failures.

"Taking bad chips and teaching them to be good, that's a huge advantage," said Steven Hillenius, SRC's vice president.

SRC also is devoting resources to figuring out what kind of device or design will replace today's transistor. The semiconductor industry estimates that by 2020, engineers will reach the limit of how much they can shrink transistors without compromising the devices' ability to operate.

With SRC taking care of basic research, companies can then spend the bulk of their research dollars on turning SRC's pre-competitive discoveries into proprietary advantages. "Each company has their own unique strategy," Sumney said. "We want to be an indispensable component of each of those strategies."