HALF MOON BAY, CA  -- Collaboration through joint development projects will be necessary for the semiconductor industry to economically deliver advanced materials required for next generation devices, according to speakers at the SEMI Strategic Materials Conference.

In the past it has taken from 15 to 20 years from pure research to high volume manufacturing, according to SMC keynoter Ken David, director of the components research group at Intel Corp. "We don't have that kind of time anymore," he said. "In order to realize the potential of new materials, we must shorten the time between discovery and implementation," and that means collaboration between device makers and equipment and materials suppliers, according to David.

Innovation in materials provides a path to extend Moore's Law well into the next decade through what he called "equivalent scaling" -- defined as geometric scaling assisted by innovation.

Every major introduction of a new material into the semiconductor process has been met by significant challenges. For example, the shift from aluminum to copper interconnects forced Intel to change the way it designed wafer fabs to avoid possible copper contamination, said David. A second example, the introduction of low-k dielectrics, was "one of the most surprising and difficult transitions," he said. "There have been a lot of missteps [in low-k]. We all learned the hard way," he said.

During a panel discussion Gene Banucci, chairman of ATMI, said it was essential for a materials company to collaborate when developing new processes otherwise the material was "going down a big dark hole." For its part, ATMI has about 30 joint development projects going on at any one time. "We have to work with other people to get things done," said Banucci.

The sharing of intellectual property arising from collaborative efforts was a hot topic for discussion. Gary Dauser, program director for IBM's intellectual property and licensing organization, said the simplest method was the "yours, mine and ours" approach. "IP that you develop is yours but I'm licensed to it; IP I develop is mine but you're licensed to it; and things we jointly develop...we can do with it what we want," he said. Dauser said the issue becomes more complex when the partners want to extend the IP usage to projects involving companies outside of the original joint development partners.

Jerry Coder, president of IC fabrication materials for DuPont Electronic Technologies, pointed out that the economics of developing materials for advanced semiconductor processes are getting poorer and poorer in terms of the return on investment. As a result, some major chemical companies have withdrawn from the semiconductor industry. "It is a problem for all of us. We need to find a model that is a win-win-win situation for all parties that are involved in that collaboration," he said.

John Poate, chief technology officer for Axcelis Technologies, said the materials used by the industry over the past 40 to 50 years have essentially been "gifts from God", and that it will get a lot harder from here. New materials, such as high-k dielectrics, are only just being explored and their successful use will require a lot more understanding, he said.

"The big IP winners will be the organizations or companies who understand the interplay between circuit design, the processing and these new materials and new structures so you can leverage what you have got," said Poate.

While collaboration is essential, speakers generally agreed it was not feasible to bundle IP developed jointly by equipment and materials companies into a single entity. Richard Faubert, president and CEO of AmberWave Systems, said the business models and infrastructure requirements for materials and equipment companies were so different that it would be nearly impossible to maintain the two under the one roof. "There are very different sets of competencies required to be a premium supplier in both places at the same time," he said.