Journal of Vacuum Science & Technology B, Vol.18, No.4, 1874-1885, 2000
Determination of coupled acid catalysis-diffusion processes in a positive-tone chemically amplified photoresist
Acid diffusion during postexposure baking is viewed to be a limiting factor in the extension of lithography using chemically amplified resists to formation of nanoscale features. Quantification of thermally activated reaction-diffusion kinetics in these materials is therefore an important step in understanding the extendability of this class of resist systems. Previous investigations have addressed this issue, however there is poor agreement among them, and too few data exist in the literature to allow the systematics of the effect of polymer, photoacid generator, added base or other resist components on the diffusion process to be understood. We describe in this article a combined experimental and modeling protocol that is designed to elucidate the chemistry and physics of the reaction-diffusion process. Because it is physically based, not phenomenological, it provides a means of developing a set of predictive, mutually comparable data. that will allow new insights to be developed into the nanoscale behavior of chemically amplified resist materials. We apply the protocol to a p-t-butyloxycarbonyloxystyrene/bis(t-butylphenyl)iodonium perfluorobutanesulfonate positive-tone photoresist system. The resulting kinetics measurements show that diffusion is environment sensitive and describable with two limiting diffusion coefficients. The Arrhenius parameters for the coefficients in p-t-butyloxycarbonyloxystyrene are D-0=1.9x10(8) cm(2)/s and E-a = 36.5 kcal/mol; those for diffusion in the deprotected polymer product p-hydroxystyrene are D-0 = 9x10(8) cm(2)/s and E-a = 22.1 kcal/mol. The coefficients are much smaller than previously reported, resulting in a very slow diffusion rate. The model indicates that the considerable image spreading observed during the postexposure bake process is attributable primarily to the efficiency of the catalytic chemistry. Our results suggest that numerical models currently used for prediction of imaging in chemically amplified resists may require refinement in order to be useful for feature sizes below 100 nm and for new classes of resist systems.