Acid Transport in Chemically Amplified Photoresists
MetadataShow full item record
Chemically amplified resists (CARs) are a class of lithographic materials that enable high-throughput semiconductor patterning. CARs are comprised of a glassy polymer resin (reactant) loaded with a photoacid generator (inactive catalyst). Patterns are formed by locally activating a strong acid catalyst with light, and then heating the film to promote catalyst diffusion coupled to polymer deprotection. While CARs have been studied for more than 40 years, there are no quantitative models that predict spatial extent of reaction with nanoscale resolution. This poses a significant roadblock for materials design and optimization, as next generation manufacturing processes will target sub-10 nm feature sizes. We studied reaction kinetics in a model CAR using infrared absorbance spectroscopy and spatially-resolved stochastic simulations. CAR formulas were based on poly(4-hydroxystyrene-co-tertbutyl acrylate) resin, onium salt photoacid generator, and an inert plasticizer. Deprotection rates were measured as a function of catalyst loading, plasticizer loading, and temperature (always below the polymer’s glass transition). Experimental data were interpreted with a simple and efficient model based on anomalous acid diffusion and a phenomenological second-order acid loss. This model predicted key aspects of the macroscopic deprotection rates, such as fast reaction at short times, slow reaction at long times, and a nonlinear dependence on acid loading. Reducing the size of the acid-counterion pair, adding an inert plasticizer, or increasing the temperature will enhance acid transport rates and reduce the anomalous character. These behaviors suggest that acid diffusion is coupled to dynamical properties of the glassy polymer resin. To complement analysis of bulk kinetics, we simulated nanopattern formation using the anomalous acid transport model, and then compared predictions with experimental line widths. The simulations include a spatial distribution of acid catalyst that reflects the exposure statistics in electron beam lithography experiments. However, while experiments include a pattern development step that dissolves the reacted polymer, the simulations do not yet have this module. Nevertheless, the predicted and measured pattern dimensions are in qualitative agreement, suggesting that lithographic resolution in CARs might be predicted from simple spectroscopy measurements coupled to spatially-resolved simulations.