The Fine Print

When Intel speaks, people are generally interested. And when the überchipmaker says it won’t be using EUV lithography to introduce 22 nm technology into its chips, people take notice. And so, like many others, I was quick to click on yesterday’s headline, “Intel: ‘EUV Facts Don’t Add Up’ for 22 nm in 2011.” It’s somewhat early still, but so far the article has garnered close to 3× as many page views as the next most popular article on our site this week (a story about Hynix cutting its capital spending, in case you’re interested). I’m not surprised, and actually went to our web analytics tool today specifically to confirm my suspicions.

In fact, the interview that my colleague David Lammers did with Senior Intel Fellow Mark Bohr was pretty interesting. But the statement from Bohr that Intel will not be able to make use of EUV by 2011 should not be at all surprising.

EUV lithography still faces several daunting challenges, including achieving defect-free masks, high-power sources with reasonable lifetimes, and improved resist sensitivity. With all that needs to be done, nobody thinks that EUV will be ready for volume production by 2011, as Bohr points out.

But what I do continue to find surprising is that Intel — as well as other key chipmakers — is still counting on EUV to come through later in the 22 nm generation, “or maybe 16 nm.” Maybe 16 nm? Aren’t we getting to a point, then, when standard CMOS scaling really comes into question? If we’re really talking about EUV not even being put into volume production until 16 nm, how long are we then even going to be able to keep it around?

Granted, the timeframe in which Intel introduces what it calls its 22 or 16 nm generation is not exactly on par with the timeframe or even physical dimensions of other chipmakers. But the industry seems to be nonetheless clinging to an awful lot of faith that EUV is eventually going to come through for them.

After reading yesterday’s news item, one reader commented, “More effort should be on alternatives to EUV than on resuscitating it.” If Intel can make it to 22 nm by extending 193 nm immersion lithography with “various techniques,” easier alternatives ought to be able to take the industry as far down traditional scaling paths as possible.

Of course, there are no easy answers — EUV, or otherwise. The EUV community has introduced another conference to the standard mix, with the International Workshop on EUV Lithography coming up in June in Maui, in addition to the EUV Symposium in the fall, this year in Lake Tahoe, Calif. What I’ve heard about the newly added EUV workshop was that Vivek Bakshi was trying to put something together that was more workshop-like (thus the name). I’m interested to see how it turns out. Obviously, there is much to be done still on EUV, and the more the industry can cooperate, the better.