EUV lithography is plagued with several problems, and these need to be solved before EUVL can be adopted by the chip manufacturing industry. This was made especially clear in Wednesday morning’s keynote speech by lithography scientist and Intel fellow Sam Sivakumar at the EUV Litho workshop in Honolulu, Hawaii. Sivakumar said that Intel is proceeding down the old ITRS roadmap in the traditional “new every two” fashion. Later this year they will release their first chips based on the 32-nm process technology, which puts them on track for 22-nm commercially available chips in 2011. Sivakumar was also quite frank about the 15-nm chips, which he says are slated for 2013. And yet despite that certainty, he would not be pinned down about what methods Intel would use to print these chips. Intel is hedging its bets: one “lighter” flavor of double patterning lithography creates the 32-nm chips. When it’s time to crank out 22-nm chips, Intel will be knee-deep in design rules. Sivakumar speculated that by the time the 15-nm process technology becomes a necessity it’s an even bet between EUV and continuing to torture the equipment and the engineers into resolving those infinitesimal line widths.
That evidently suprised Chris Mack, because he got up at the end of the keynote and pressed Sivakumar on whether they were maybe even one percent more in favor of one of the technologies at this point. Sivakumar wouldn’t budge.
On an unrelated note, I asked Mack: If someone had a gun to your head, and asked you to bet on the next-generation technologies: EUV lithography or E-beam Lithography (the increasingly attention-getting technology coming out the Netherlands), what would be your bet?
“I’d bet on the gun,” said Mack.