Using mesoscale simulation to explore photoresist line edge roughness

Computer simulators are ideal tools to study complex process spaces, but current lithography simulators are based on empirically-derived continuum approximations and thus are unsuited for investigating properties like line edge roughness (LER) because they do not incorporate molecular level details. A "mesoscale" simulation is described that enables molecular level effects to be captured. This technique is a compromise between accurate, but slow, atomic-level simulations and the less accurate, but fast, continuum models. The modeling of stochastic processes that lead to LER is enabled via use of Monte Carlo techniques. Mesoscale simulation was used to study the effects of added base quencher to overall photoresist performance. Simulations of acid/base kinetics with quencher loadings ranging from 0 to 20% show good qualitative agreement with experimental data. Results show that decreasing aerial image quality increases the root-mean-square (RMS) roughness, whereas increasing base quencher loading improves LER, up to approximately 50% base. A mechanism that explains line edge roughness stemming from acid gradients is proposed. This mechanism is supported by simulations showing that the catalytic chain length varies inversely with acid concentration. Simulation results show that base effectively limits the influence of acid in low concentration regions. A critical drawback of using base additives is significantly reduced photospeed.