Abstract

The oxidation of o-xylene and/or naphthalene to phthalic anhydride is one of the important industrial processes based on catalytic selective oxidation reactions. Vanadia - titania catalysts have been used in the industrial phthalic anyhdride process for the last 50 years. The operation parameters like the temperature range of operation, reactor inlet pressures, contact times, o-xylene loadings, etc. were constantly improved during this period of continuous process optimization so as to optimize catalyst performance and increase its life time. However, a fundamental understanding of the mutual interaction of the rather complex reaction network and the catalyst formulation is still missing. Recently, a detailed study of by-product formation as function of process conditions allowed us to develop a novel, improved reaction scheme for the catalytic oxidation of o-xylene [1]. Based on this understanding, a detailed investigation was conducted for the first time of the by-product formation under varying operation conditions and as a function of the active mass variation exploiting high-throughput, as well as bench scales reactors. This high-throughput testing allowed us to relate reaction kinetics to novel catalyst formulations.

Keywords: Artificial neural network, kinetics, phthalic anhydride, reaction network, vanadia-titania catalysts, oxidation reactions, high-throughput, o-xylene, feedstock, multilayer catalysts, isothermal operation, homogeneity