SI's Take on Semicon West

There was lots to be said about extreme ultraviolet (EUV) lithography at SEMICON West — some of it good, some of it not so good. On the one hand, Cymer announced that it would be ASML's EUV source supplier going forward. Along with that announcement, Cymer detailed the advances made in its laser-produced plasma (LPP) source design (Cymer switched from DPP to LPP three years ago). Cymer has managed to get IF power up to 50 W, which is significant progress in output power. And the lithography source manufacturer promises to reach 100 W by the end of the year (and, according to Cymer's Nigel Farrar, the company has hit every one of its marks since setting its roadmap at the beginning of last year). That's downright close to the target the industry has been aiming for to achieve production throughput on EUV litho tools. But it is, of course, a moving target. Setbacks in resist dose and the number of mirrors needed in the EUV system threaten to keep boosting the power needed from the source. The power requirement could actually go up by a factor of two or three, Farrar said.

There's still a lot of murmuring from the rest of the industry about debris mitigation with these EUV tools. But Cymer has made significant progress there too, enabling a collector lifetime of about a year. Nonetheless, Cymer has its work cut out for it now. The next 18 months will show whether the engineers can really pull it all off and meet the schedule ASML has in place.

Certainly, plenty of people in the industry are not convinced that EUV lithography is the right way to go. They are latching on to double patterning as a way to get them to 32 nm (since there's little doubt that EUV won't be the answer at that node, anyway). EUV has become an Intel-led political battle, with many employees jumping ship because of it — whether immersing themselves in Sematech activities or full-out quitting their current companies, they are finding ways to get away from the EUV development work.

Not surprisingly, the folks at nanoimprint lithography toolmaker Molecular Imprints do not speak highly of EUV. I met with them at SEMICON West, where they detailed many of the difficulties EUV is facing. The meeting was arranged by John Doering, who said he left ASML for Molecular Imprints specifically because of ASML's dogged focus on EUV lithography, and the unwillingness to spend any time or money exploring what he saw as the more sensible nanoimprint approach. Mark Melliar-Smith, CEO of Molecular Imprints, wondered aloud why chipmakers would want to use a technology that wasn't likely to take them very far. "I don't think EUV can get down to 10 nm," he said.

Nanoimprint lithography has its own hurdles to overcome, of course, but that's a story for another day...