I don't think you understand how wavelength affects lithography... the manufacturers are currently at 193 in wavelength yet are patterning 28nm (and beyond) now you say 12.5nm (which is incorrect actually) will be one node? Once EUV is up and working it will literally be able to pattern all the way down to 1 (maybe beyond?)
i do not think that anyone can predict whether the technology will fail or succeed with full surety. For ASML, emersion bet succeeded but EUV has failed to deliver. Still, the industry hopes are riding on it fortunately or unfortunately.
I find it difficult to correlate the record sales of ASML with the dismal progress of EUV, by the correlation ASML should be at the bottom and not at the top. The success of ASML should be judged by the market captured by the emersion tools and one of a kind technology provider in the market.
There's a reason why ASML is basically the lone EUV tool vendor. I believe that Nikon had/has the ability to be at least a close follower to ASML in EUV litho. But I think that they foresaw that the significant investment required for EUV technology development would be difficult to recoup. It was a business decision. They didn't drink the EUV litho kool aid.
EUV litho isn’t the only path to higher chip performance. New materials (FD-SOI), new architectures (3D/TSV, FinFETs) not to mention advances on the EDA front all have the potential to drive higher levels of performance and lower costs or at least bend the cost curve. Each of these approaches have their own technical hurdles and cost issues, but is EUV really that far ahead of the pack and any more “real”? And as far as Moore’s Law is concerned, the huge costs associated with EUV at the production-level kills the whole cost-side of the Moore’s Law equation. If it becomes a one-node solution, then does it really make economic sense? I’d love to see an economic report on the projected payback period of EUV once all of the dev costs are calculated.
EUV litho is still a problem in my mind - and not a sure thing. At least not for vast majority of semi companies. Yes, my opinion does go against industry consensus that EUV is the default next gen litho technology, but as has been pointed out by others, it will likely not be ready until some sub 11 nm design node. At that point the current wavelength (12.5 nm) will render it a single node solution. So EUV needs to migrate to 6.7 nm wavelength soon. That puts even more pressure on the source and resist vendors who currently have most of their eggs in the 12.5 nm basket. Plus you still have the reticle fabrication and inspection issues and the over $100 million price per tool. I believe that the technical challenges are still great. They will eventually be overcome, but Moore's law will have been broken to bits by that time. So EUV will become a relatively niche technology for a handful of companies who have the need for very high volumes that they can sell with high cost margins.
That consensus may have been last year, when people were still sure about power throughput.
The 14 nm logic in 2013 and 10 nm logic in 2015 will happen without EUV for sure. And memory will also have to move on during this time without EUV. Not enough systems available for everybody.
They'll learn to live without it.
The consensus in our industry is that EUV will happen. The question is when and at what cost and throughput.
Quad patterning with immersion tools will probably continue till 11nm with double spacer patterning... EUV may have an opening after that, in combination with or in parallel to 3D-type scaling
A Book For All Reasons Bernard Cole1 Comment Robert Oshana's recent book "Software Engineering for Embedded Systems (Newnes/Elsevier)," written and edited with Mark Kraeling, is a 'book for all reasons.' At almost 1,200 pages, it ...