by Debra Vogler, Senior Technical Editor
In a fab-lite/process-lite business environment, it seems logical to expect that consortia will have even more power and responsibility as more companies rely on a shared cost model for process and materials R&D.  In addition to the usual consortia membership rolls of IDMs, foundries, and equipment/materials suppliers, IMEC’s director of strategic program partnerships, Lode Lauwers, says he expects memory manufacturers, and more recently, fab-lite converts, to add their own needs or “flavors” to consortia research programs.
A speaker on the Monday panel titled “The Economics of the Semiconductor Industry Ecosystem,” Lauwers explained in an interview that fab-lite companies need enough information to make the right decisions as to which of their processes should be farmed out to external foundries and which should be made in their own fabs. The fab-lites will also need the information provided by those participating in consortia to further tune their processes — even those they send out.
“Besides the fab-lite companies, we see the memory manufacturers playing a more prominent role [in consortia] than in the past,” said Lauwers, noting that memory firms are leading in technology research and there are commonalities in searching for new materials. Additionally, memory and DRAM manufacturers are expected to gain further improvement through scaling so they can use the knowledge and learning that consortia have achieved in logic scaling. “At IMEC, we won’t work on a specific DRAM cell — we will leave that to the memory companies — but we will address high-k solutions for DRAM cells, just as we are doing for logic.”
Perhaps the most challenging aspect for consortia in terms of maneuvering in a fab-lite/process-lite environment may be how to work with equipment and materials suppliers who happen to be competitors. Asked if suppliers might end up aligned with different consortia, Lauwers pointed out that high investment costs drove IMEC’s strategic alignment with ASML for lithography, but in other front-end and back-end areas IMEC does work with competing suppliers. “We try to make the research in those areas complementary because it’s not so easy to have people [on staff] working on the same topic with competing suppliers/tools,” he said.
Despite the ASML example, Lauwers reiterated that IMEC is not out to make exclusive deals with suppliers. “Our equipment decisions are based on our members’ requests, the capabilities of tools, and certain techniques,” and the consortium wants to be sensitive to concerns. “We don’t want to confront suppliers in one area and then bring in a competitor’s tool in the same area — it becomes very difficult to control the IP,” especially if the same people are working on both tools, he said. “It’s a balance between having the right capabilities on board and managing the IP.”
Lauwers also pointed out that care must be used in avoiding the problem of having a specific research result developed on one supplier’s tool, and then copied on a tool from a competitor who did not pay for the development research. “We cannot just say we’ll put tools next to each other — we have to enter into discussions,” he said.
Summarizing the changes for consortia he sees in the near term, Lauwers believes that there will be fewer players doing aggressive scaling overall, and programs will be driven by a combination of inputs from leading IDMs, leading foundries, memory companies, fab-lites, and equipment suppliers. These five groups will bring with them their inputs and requests that will form the ideal program for technology research.
Other changes Lauwers foresees are an expansion of 3D integration, and a growing interest in packaging technology. In particular, he pointed to IMEC’s “More than Moore” program, which explores what can be done in addition to scaling by adding more functionality on a chip, such as by combining CMOS with MEMS processing. — D.V.
 For more on this theme, see: “The progeny of chip commoditization: Fab-lite AND process-lite” (WaferNEWS, April 3, 2007); and “Virtually vertical IDMs: Bridging the gap between “fab-lite” and foundry” (WaferNEWS, April 17, 2007).