Statistical analysis of the impact of 2D reticle variability on wafer variability in advanced EUV nodes using large-scale Monte Carlo simulations

Transfer of reticle variability to wafer variability has received extensive attention in previous studies. In general, it is well understood that reticle contributions depend on many factors including reticle fabrication quality and lithography:design co-optimization (pitch, k1 factor, source:mask optimization, MEEF, etc.) In this paper, we present an analysis of reticle variability impacts based on large-scale (10^5 - 10^6 patterns) simulation of masks with programmed variability. Simulated variability is introduced using Monte Carlo approaches and allows us to probe impacts of different sources of variation: • Pattern size variation related to dose-like variation during reticle manufacturing. • Pattern position variation related to shot placement variation during reticle manufacturing. • Pattern roughness (expressed in terms of amplitude and correlation length) related to reticle manufacturing and resist stochastics. Carrying out simulations in this way allows us to measure variability distributions and their low-probability impacts at approximately a 5-sigma level. We apply this simulation method to various test cases relevant to N3 and N2 node scaling including DRAM, SRAM, and logic layers.