Japanese companies collaborate for semiconductor development

04/01/2006

Collaboration appears to be a sign of the times and may impact future cleanrooms

By Hank Hogan

In February, three Japanese companies-NEC Electronics Corp., Sony Corp., and Toshiba Corp.-announced they were teaming up for process development of the 45-nanometer semiconductor node. For cleanrooms and the contamination-control industry, it was another indication that the future is shrinking in more ways than one. Along with finer geometries, there’s likely to be an accompanying shrink in the number of semiconductor players and a narrowing of the market because of such joint development efforts.

In the case of the Japanese companies, the trio will be working together on process research and development (R&D) at Toshiba’s Advanced Microelectronics Center in Yokohama. In a statement, the three said they were doing this to raise development efficiency and accelerate the pace of development.

As a result, three independent process development efforts, which could have resulted in different specifications for process recipes and tools, will now effectively be one. While such collaborations aren’t new, the trend is accelerating. “There’s definitely more cooperation with every node,” says G. Dan Hutcheson, CEO of the chip manufacturing market analysis firm VLSI Research, Inc. (Santa Clara, CA). “It’s being forced by this long-term escalation in R&D costs.”

As proof of this increasing cooperation, Hutcheson points to Samsung. After years of going it alone, the Korean chip maker, one of the largest in the world, joined Sematech’s group manufacturing R&D effort in March 2005. As for R&D costs, VLSI Research’s figures show those increased from about $5.4 billion worldwide for the 1500 nm node in 1984 to $22.5 billion for the 350 nm node a decade later, and then topped $70 billion for the 65 nm node in 2005. The final cost for the 45 nm node will be higher still.

These R&D expenses have spiked upward since 2000. That’s when feature sizes moved below 100 nm, and well-established rules governing how to scale different aspects of semiconductor components for smaller feature sizes broke down. To overcome this, the industry has had to bring new materials into the process and do much more transistor engineering. Only then were manufacturers able to mass-produce chips with an economically acceptable yield.

One result has been a tighter coupling between the design of a circuit and its manufacturing. Consequently, when a joint process-development collaboration establishes that a certain tool will be used or a particular contamination-control procedure followed, it becomes difficult for any of the individual companies to change direction later. “It’s all interconnected now. It’s not like each tool can be dropped in and everything works,” says Hutcheson.

Michael O’Halloran, director of technology at the cleanroom construction firm CH2M HILL IDC (Portland, OR), also has witnessed increasing collaboration in the semiconductor industry. He believes that procedures, processes, and vendors that are selected by these joint development projects have a leg up on those that aren’t.

For the most part, O’Halloran believes that the impact of increasing cooperation has fallen, and will continue to fall, on tool makers, since the corresponding products and vendors are closer to the manufacturing process than are cleanrooms or their construction firms. Nonetheless, he does see some fallout in IDC’s area of expertise. “Companies that already are focused on the semiconductor industry and dependent on it will have a chance of surviving. Those that aren’t will probably wander off and do other things because the initial hurdle is too high,” he says.

There is a contradictory piece of evidence and perhaps a bright lining in all of this, though, for cleanrooms. VLSI Research’s Hutcheson has tracked the semiconductor industry for years, and he’s found that construction activities don’t always follow the course suggested by a strictly economic analysis. Such studies would probably have concluded, for example, that there was no need to build semiconductor cleanrooms in either Taiwan or Korea. As Hutcheson says, “There’s systematically been a lot more cleanrooms built than one would suspect.”