Blaming photons at the most advanced nodes doesn’t provide the whole picture of what’s happening.
The line edge roughness of a chemically amplified resist ultimately depends on photoacid generation and the deprotection of the resist’s base monomers. Photons absorbed by the resist simply trigger a chain of events leading to deprotection. In EUV resists in particular, the absorbed photon actually ionizes a monomer, setting off a chain of secondary electron emissions that, in turn, trigger photoacid generation. The diffusion distance of the photoacid molecules defines the “deprotection blur.”
At first glance, it appears that increasing the deprotection blur allows more photons to contribute to each feature. In a shot noise-driven roughness model, this would be a good thing. However, as the diffusion distance becomes large relative to the feature size, there is less chemical contrast between exposed and unexposed areas. Moreover, increasing blur affects the ultimate resolution of the resist. Similarly, if the fraction of base monomers in the resist increases, more photoacid will be needed to achieve the same degree of deprotection. Again, it is the photoacid distribution, not the shot noise, that ultimately defines the resist’s line edge roughness.
Unfortunately, attributing line edge roughness to the photoacid generator distribution does not bring the industry any closer to a roughness solution. Shot noise poses an intractable problem because the small number of EUV photons in any given area will be randomly distributed, not necessarily conforming to the outline of the desired feature. The same, however, is true of photoacid generator molecules, of resist base monomers, and of most other contributing factors. As Chris Mack has said, solving line edge roughness will require non-random resist behavior.