There are times in life when you understand that you are not so young anymore.
To me, it happens when I think about EUV technology.
I still remember very well when EUV was started to be heralded like the next generation breakthrough technology for lithography steppers.
I was an avid technology reader at the time and I first heard about EUV around 1995 or 1996, if I remember well.
Few years later, I remember the first discussions about when EUV would have been ready for production, I remember few analysts claiming EUV would have been ready for the 65nm node.
Well, this would have meant to have a production-ready tool by 2005.
Needless, to say, EUV never made it for an insertion at the 65nm node, then skipped the 45nm node as well, the 32nm node, the 22nm, the 16nm and now ASML is talking about alate insertion at the 10nm node, but most analysts are skeptical that the technology will be ready on time
I remember that few years after I started to hear talks about EUV, another technology was coming out as the most promising NGL (next generation lithography) candidate: 157nm.
157nm lithography was in some ways less revolutionary than EUV, it was the logical step in the traditional evolution of lithography, from using 248nm wavelengths lasers, to 193nm finally to 157nm
157nm lithography actually never made it into production as alternatives like immersion and multi-patterning lithography became mainstream.
So what about EUV?
EUB is still in the game, so much in the game that recently Intel has invested some $2.1billion in ASML and in return ASML has bought Cymer, a light-source maker specialized in light-sources for EUV.
At a party at the Belgian Embassy here in Tokyo, after the annual conference of IMEC at the New Otani Hotel, I approached a senior researcher from IMEC and asked why, despite the numerous delays of EUV, still the major players in the industry were pouring billions in this yet to be completely proven technology.
His answer was plain and honest: “There are no other options”
And his comment proved to be true: all the technologies that were hoped to become a potential replacement in case of EUV failure become players in other markets or simply disappeared.
Multiple beam E-beam technology, proposed by Mapper who got funded by TSMC, was assumed to become a player at the end of last decade but the hopes failed to materialize.
Nanoimprint technology is alive, kicking and growing, but not for IC manufacturing (despite Molecular Imprints` s claims to have a soon-to-be-ready IC-patterning capable tool). The technology has found a vast interest in the bio, film-replica and other markets.
Multiple patterning and immersion lithography have been heavily used to extend the life of conventional 193nm lithography so far, but costs are increasing exponentially at every node, and multiple patterning, once a synonymous of double-patterning, is now meaning triple, quadruple or even quintuple-patterning.
Direct self-assembly is promising, but still very much at the R&D level.
So what is left?
Not much, I would say.
Despite the claims of Peter Wennink, the new CEO of ASML, that EUV will, finally, be production-ready soon, we come to know that the technology will reach improved standards only by 2017, this is, definitely late for the 10n node and possibly late for the 7nm node.
While I do believe that EUV may be ready for mass-production before the end of the decade, the main issue would be if EUV will be ever seen as the savior of Moore` s Law or if, instead, will be relegated as a complementary technology with the mainstream production ecosystem moving to alternative solutions like 3-D stacking, for example.