Molecular-scale kinetic Monte Carlo simulation of pattern formation in photoresist materials for EUV nanolithography

A kinetic Monte Carlo (KMC) simulation tool for modeling the pattern formation process in photoresist materials for extreme ultraviolet (photon energy 92 eV) nanolithography is presented. The availability of such a tool should support the progress toward novel materials and experimental procedures that lead to an improved pattern resolution. The molecular-scale simulations describe the process in a stochastic and mechanistic manner and include the excitation of high-energy electrons upon light absorption, the creation of a charged-particle cloud, electron-induced chemical degradation of the photoresist molecules, the resulting bond formation between neighboring degraded molecules, and a chemical development step after which a pattern of the remaining non-dissolved molecules is obtained. The method is applied to the application-relevant class of Sn-oxocore photoresist materials and uses their known electronic structure and optical electron energy loss function. The validity of the approach is tested by comparing measured and simulated total electron yield spectra and photoelectron spectra. A demonstration of the method is given by calculating the dose and pitch dependent average shape and stochastic variability (line edge roughness) of line patterns that are obtained for rectangular and sine-wave illumination, assuming various scenarios that determine how molecular-scale degradation will lead to bond creation. We show how from these simulations the ultimate pattern resolution can be deduced. The findings are analyzed systematically using results of KMC simulations that reveal the size of the cloud of degraded molecules around a point of absorption (blur length) and that further reveal the sensitivity to uniform illumination (contrast curves), and using percolation theory. We find that KMC modeling captures the consequences of the strong gradients in the density of degraded molecules and of the stochasticity of the patterning process that simplified models do not include, leading to a significantly improved view of the final pattern quality.