Originally, it was hoped EUV would be ready in time for the shrink to the 22nm node but getting these extremely complicated machines ready for mass production turned out to be much harder than ASML expected when it starting working on this new technology at the start of the millennium.
The first EUV prototype machine was ready in 2006 but several generations later the company is still in a race against the clock to ensure its clients, which include Intel, TSMC, GlobalFoundries and Samsung, will be able to kick off EUV mass production towards the end of 2018 or early 2019.
The chip makers all have a close cooperation with ASML because this Dutch company is the only one in the world that manufactures cutting-edge chip production equipment. Japanese firms Nikon and Canon also participate in the lithography market but have steadily lost marketshare to ASML.
Wennink claims EUV arrives just in time for the 7nm node as the currently used immersion lithography is hitting the limits of what is technically and economically possible. This is one of the reasons why companies are struggling right now with new nodes and why Intel delayed its 10nm node.
Chip makers were able to stretch the lifetime of immersion by adopting multiple patterning techniques, but this had an adverse impact on yields and costs. Intel hopes that with the introduction of EUV it will be able to get back to its tick-tock cadence.
The ASML CEO adds that EUV will make the 7nm, 5nm and 3nm process nodes cost-effective. Furthermore, he also reveals that current .33 NA (numerical aperture) EUV will likely be good for two nodes, before switching to High NA with a numerical aperture of over .5, which in turn should be good for another two to three nodes.
Obviously, significant technical challenges remain but ASML expects EUV will have a horizon of beyond the year 2030. Some clients reportedly have roadmaps that reveal a path to the 1.5nm node with EUV.