In extreme ultraviolet lithography, multilayer roughness effects are a key contributor to mask-induced pattern roughness1. In particular, replicated roughness from the mask substrate results in a spatially dependent phase error that ultimately manifests as aerial image roughness at the wafer. In this paper, we utilize Monte Carlo methods to study the impact of multilayer roughness on the printing of two-dimensional patterns. We focus on the impact of multilayer roughness on both TaN absorber masks as well as attenuated phase shift masks, comparing roughness impact on each architecture at both 0.33 and 0.55 NA EUV lithography. We examine whether the modeled multilayer roughness explains observed patterning phenomena, with a focus on what level of remedy would be required for multilayer roughness effects to become negligible components of the patterning error budget.
|