English: Shot noise affects the diameter of an individual feature printed by EUV. If the inner and outer dashed circles indicate 25 nm and 27 nm diameters, respectively, the difference in area is 82 nm². A dose of 50 mJ/cm² is 34 EUV photons per square nm. Therefore, about 2800 photons are expected in the region of variation for this case. Poisson distribution for shot noise^[1] indicates a standard deviation of 53 photons (~2%). Six standard deviations widens the printed diameter variation range to ~11%. This assumes 100% photon absorption. More realistically, 25% absorption would double this noise level (to ~23%).

The secondary electron density in the region of variation is even lower, with the electrons likely to migrate outside this 2 nm wide region. This significantly enhances the noise component from the secondary electron contribution. For ArF (193 nm) lithography, the shot noise itself is less severe due to more photons used for larger feature edge areas, even though a smaller fraction of photons are absorbed. For example, if the inner and outer diameters are 36 nm and 40 nm respectively, and the dose is 35 mJ/cm², with 4% absorption, ~3300 photons are absorbed in the region of variation, with six standard deviations being less than 11%.

Hence, EUV is at a disadvantage compared to immersion lithography as far as resolution limits from stochastic dose fluctuations are concerned.^[2] This prevents EUV from demonstrating its more widely recognized advantage in diffraction-limited imaging.