You are currently viewing SemiWiki as a guest which gives you limited access to the site. To view blog comments and experience other SemiWiki features you must be a registered member. Registration is fast, simple, and absolutely free so please, join our community today!
The 157nm immersion approach got us to sub-40nm lithography, however starting at sub-28nm we had to start using multi-patterning, or multiple masks per layer. EUV has a 13.5 nm wavelength and this allows the industry to do many of the critical layers in 11nm and smaller nodes. Mask costs are high, and EUV costs are even higher, so only the highest volume ICs can afford this new technology.
157nm would have obtained 1 more generation at best, assuming everything that works at 193 would work just as well with 157. In practice, it doesn't. The materials that are transparent to 157 are much more restricted than 193, and do not include water. The lasers are harder to make and maintain, the lens materials not as good. LELE patterning could be redone, but would need new chemistry. And all of that just for one generation? EUV is a massive amount of work but seems possible to extend for several generations. But actually, 157 was bypassed even by 193, as the 193 toolset simply matured faster than anyone could solve the problems to get 157 accepted.
Actually 157nm non-immersion lithography was already killed by 193nm immersion because of the engineering problems for 157nm. Also I don't think you will find any liquid with right properties to be used as immersion fluid (viscosity, needed purity, etc) and be transparent enough for 157nm light.
Could not be extended by immersion like 193nm was because the materials were exotic....for example the liquid needed to be something like decane which is a hydrocarbon. Now try to recirculate decane (gasoline) between a lens and a wafer (with resist on it) [Safety and fire hazard in a multimillion dollar tool was not something the industry was looking forward to]. The resist was a fluoropolymer (for transparency) and fluoropolymers are expensive (even for non-immersion). Of course as stated above the lenses were CaF2 (also expensive)....that could have been done. Intel pulled the plug (got out) and that was the end. So 193nm and 193nm immersion was a more than one generation solution. Photoresists for 193nm and 193nm Immersion were almost identical (once you added a protective layer on the resist)...a thin fluoropolymer....which more or less was already available (because of other fluoropolymers already in use). So the industry had all the tools necessary to move forward quickly from 193nm to 193nm immersion. Water is used in 193nm immersion and so not many issues.
193nm immersion already had more resolution than 157nm non-immersion. There is no good immersion liquid for 157nm immersion so it wasn't considered. You don't just need a transparent liquid, you need a high refractive index at 157nm. Even If you had a 157nm liquid comparable to water for 193nm (which doesn't exist) , this would still only give a 20% resolution improvement over 193nm immersion. This isn't much for a whole new exposure technology and nowhere near the promised improvement of EUV.
Many have answered already, correctly indicating the 157 nm material issues.
EUV offered the promise of a return to higher 'k1', which would have been 'easier', more conventional lithography. This is calculated by feature size x NA/wavelength, and is higher with the much shorter wavelength of EUV.
What turned out to be profoundly difficult was the use of reflective optics and the low EUV photon number density, leading to shot noise and subsequent stochastic issues.