|
ISMI Symposium drilldown: He who is without R&D collaborators is lost |
|
|
|
Oct 31, 2007 at 10:58 AM |
A discussion of the challenges of funding semiconductor research and development may not be a new idea, but that doesn't mean there's nothing left to say.
An R&D-related panel moderated by Dan "Don't forget to visit our Websites" Hutcheson of VLSI Research at last week's ISMI Symposium managed to be compelling and even offer a few fresh comments and viewpoints. The panelists---Larry Sumney, long-time head of the Semiconductor Research Corp.; Mike Polcari, current Sematech boss; Frank Robertson, Intel's tech and manufacturing group's external programs manager; and Paul Farrar Jr., IBM's VP of process development---provided the intellectual fodder.
Sumney, recent recipient of the National Medal of Technology, said that since the launch of SRC, its programs have graduated 6000 students (mostly PhDs) and invested nearly $1.1 billion in university research so far. He noted how the "long threads" went back for years into successful basic-research programs in circuit design and verification and high-k dielectrics. During the Q&A following the initial comments of the panel, Larry responded to a question about what the U.S. has to do to deal with a declining "talent base." In addition to bolstering and promoting science education at the primary and secondary level, he pointed out that "we have the best university system here, and we need to keep it strong," adding "it doesn't take a lot of money to make a lot of difference at the undergraduate level."
Mike Polcari recalled how in the early days of Sematech, it took months for people from competing companies to really talk to each other, but today, the sharing of manufacturing data has become commonplace in consortia and alliances. "Back in the day, that [info sharing] would've been the end of your organization," he quipped. Summarizing Sematech's role, Mike said the organization helps the industry "make data-driven decisions, build databases to allow the members to make the right choices going forward....You need to see what doesn't work as well as what does work to make decisions.... Development is like evolution---some branches don't go any further."
Dispelling somewhat the myth of Intel as a go-it-alone company, Frank Robertson noted the company's active role in Sematech and IMEC as well as its comprehensive network of external research projects at universities and national labs. "Consortia are a good value for Intel. We can look at alot of different options, focus on the good options for our internal R&D and then focus on what will work." The trick is deciding what should be external and internal research, finding the spending and effort balance in what he called "a distributed Bell Lab...distributed research capability" model. When asked about how the model worked in the development of high-k dielectrics/metal gates, Frank noted how the university work looked at the fundamental materials properties, then the "combinatorial approach" at Sematech allowed the the industry to get a good look at the options, and then Intel's own R&D kicked in and zeroed in on what would work.
Coming from a company called "the king of collaboration" for its establishment of various "fab clubs" and interindustry alliances, IBM's Paul Farrar said the problem in the early days was one of differing corporate cultures: The question was, "How do you communicate to upper management?" One of the keys, Big Blue discovered, was to create a culture that "makes all the partners look the same. About 30% of our workforce is now partners and you can't tell them apart." Paul then offered a business homily about the benefits and give-and-take of a strong partnership: "[You need to] learn to share more than you think you need to share---and you get more back. Every time you bring more to the table, you get more back." With the fab club, he noted that "we have a window into multiple, outside fabs" that wasn't there before, which has greatly increased IBM's own knowledge base of manufacturing best practices.
During the Q&A, the panel was asked what could be done next to increase collaboration and to further reduce R&D risk and pain. Their answers all looked to improving cooperation in the supply chain.
Frank said more OEM collaboration is needed: "Standardization must be extended closer to the process chamber," adding later that the industry needs to "push beyond the loadport into the tool...the value is in the chamber." Larry believes there's more cooperation among OEMs at the research level than in the past and believes "the equipment industry is seeing the benefits of working together precompetitively." In earlier comments, Paul had noted how TEL and Applied Materials "are running experiments on each others' tools" [albeit with some "recipe segregation"] and sit in together on logistical and operational meetings at Albany Nanotech.
Mike cited how increased collaboration among tool and materials suppliers represents an opportunity, while Paul said the "design efficiency space" is a potential area for cooperation, seeing a move to "an industry where designs can only do what's allowed---not just do everything except what's forbidden."
And whither Moore's Law, the quasi-metric of the industry's relentless technological/economic progress over the past 40 years? The man in corporate blue, not the guy working in Gordon's house, had an answer. "It will go on as long as you can see two nodes out," opined Paul, "and we can see 22 nm, so it's still in motion."
|
Comment by GUEST on 2007-11-01 17:03:53
Were no suppliers invited to participate? It would have been interesting to get that perspective. |
|
|
