Liquid Phase Epoxidation of Propylene to Propylene Oxide with Hydrogen Peroxide on Titanium Silicalite-1: Spatially Resolved Measurements and Numerical Simulations

The spatial profile measurement technique was applied for the first time to a liquid phase reaction to study the selective oxidation of propylene to propylene oxide with hydrogen peroxide on titanium silicalite-1 catalyst, also known as the HPPO process. A spatial profiling reactor based on minimal invasive capillary sampling measurements was used to obtain the experimental data. The reactor is able to measure spatially resolved concentration and temperature profiles at industrially relevant conditions, that is, 40 °C and 20–35 bar pressure, with propylene in the liquid phase. The reaction was carried out at different contact times of LHSV 3.0, 6.0, and 12.0 h –1, respectively, achieving high conversion and selectivity. The experimental data show that the concentration profiles of the main epoxidation reaction follows a pseudo-zeroth-order behavior. In contrast, the side product concentration profiles exhibit a nonlinear trend. The primary side products of the reaction 1-methoxy-2-propanol, 2-methoxy-1-propanol, and propylene glycol were quantified. Furthermore, hydroxacetone and methoxyacetone from the consecutive reaction of 1-methoxy-2-propanol, 2-methoxy-1-propanol, and propylene glycol with hydrogen peroxide were detected as secondary side products. Formic acid, methylformate, dipropylene glycol methyl ether isomers, and dipropylene glycol isomers were detected in trace amounts. Quantitative data for the formation of hydroxyacetone is provided for the first time. The experimental concentration profiles were simulated using an axial 1D-pseudo-homogeneous dispersion model with two kinetic models reported in the literature. Additionally, a kinetic model based on the Eley–Rideal mechanism is proposed. Upon linearization, the derived model shows a correlation of R 2 = 0.99 and R 2 = 0.98 with the linearized form of the experimental differential rate. Statistical analysis of the models in this work shows that the derived Eley–Rideal mechanism has the highest correlation with the concentration gradients measured within the catalyst bed.