You may have noticed that Intel Corp. isn't going along with the Great Copper Rush of '98. While most of the industry rushes to catch up with IBM Corp. and solve the tricky aspects of moving copper into volume production, the microprocessor giant was certain that copper metal wouldn't be needed to break the interconnect bottleneck until 2001 or 2002.
But the chip leader may be missing something here. Technology may not be the primary copper driver after all. In fact, pushing hardest at some chip makers these days to get copper into production are their marketing departments.
Indeed, no new chip production technology may be complete until it includes copper. During a recent trip to the Far East, Andy Kirkpatrick ran into a panic-stricken fab manager who was under the gun to get a copper process working ASAP.
"He was under tremendous pressure to do copper work," recalls the marketing director of GaSonics International Corp., which supplies photoresist and wafer cleaning tools. "It turns out that his marketing people were already demanding a photo of a wafer with copper."
The marketers at the Asian foundry did get their photo of a copper wafer, but they won't see real products with copper interconnects rolling off their production lines until 2000 at the earliest.
Copper processing had been primarily technology-driven until it suddenly turned into a marketing issue with IBM's initial disclosure in the fall of 1997, analyst G. Dan Hutcheson told January's International Strategy Symposium in Pebble Beach, Calif.
The low-resistance interconnect metal got an even stronger marketing push last year when Steve Jobs and Apple Computer Inc. demonstrated an upcoming system built around a 400-MHz PowerPC copper processor from IBM Microelectronics. Apple, savvy marketing company that it is, obviously sees copper as an attention grabber.
Then there was "the kid here in Silicon Valley telling his parents he wanted a computer with copper chips," recalls Martin H. Manley, director of device and process integration technology development at VLSI Technology Inc. The San Jose company now plans to offer a copper interconnect option in its new 0.15-micron process.
There is a big struggle going on now inside chip companies, according to fab equipment suppliers. There are those people who want to offer copper capabilities as soon as possible, while others aren't counting on large volumes of copper chips for several years.
"At times it's been a week-to-week change," notes Jacki Seto, director of product management for the dielectric etch operation at Lam Research Corp. in Fremont, Calif. "Right now, I get the impression companies are trying to pull in and accelerate their copper programs again."
Copper will be a complicated marketing problem for the pure-play foundries. They will need to demonstrate a copper capability with their next-generation processes that will be ready to go whenever needed by customers.
"TSMC cannot serve customers with a preference," says Ken Chen, senior manager of technology for Taiwan Semiconductor Manufacturing Co. (TSMC), who is based in the foundry's San Jose office. "Some designers will continue to favor conventional aluminum with low-k dielectrics. Others will want copper and aluminum. And still others will want all copper," he notes. "We must develop technologies for different requirements."
Perhaps marketing is now in charge after all.