Catalytic hydrogenation of p-nitrocumene in a slurry reactor

Abstract

The hydrogenation of p-nitrocumene to p-cumidine over supported palladium catalysts was investigated in a laboratory-scale slurry reactor. The primary objective was to demonstrate the methodology for development of a slurry reactor model that could predict either isothermal or nonisothermal performance using intrinsic kinetic and transport parameters that were determined from independent data and engineering correlations. Several catalysts were screened to identify a suitable one for kinetic and reaction engineering studies. Various catalyst supports, such as alumina, calcium carbonate, and activated carbon, as well as reducing agents used during the catalyst preparation, including hydrogen, sodium formate, and formaldehyde, were investigated. A 1 wt % palladium-on-alumina catalyst was identified as the preferred catalyst because it had both superior activity and selectivity. The effects of hydrogen pressure, catalyst loading, and the initial concentrations of p-nitrocumene, water, and p-cumidine on the initial rate of hydrogenation and the concentration-time profiles were also studied in a batch reactor. The initial rate data showed that both the kinetic and mass-transfer resistances were important for temperatures greater than 353 K, while the kinetic regime was controlling at lower temperatures. A Langmuir-Hinshelwood (L-H) model was proposed based on the rate data in the kinetic regime. The rate model was suitably modified to account for combined transport-kinetics resistances above 353 K. Using a basket reactor, intraparticle diffusion effects were also studied by transforming the catalyst powder used for the kinetic study into catalyst pellets. Equations for an overall catalyst effectiveness factor were derived for the L-H type rate model. The experimental data for different catalyst particles agreed well with the theoretical predictions. To verify the applicability of the kinetic model over a wide range of conditions, a slurry reactor model was also developed for both isothermal and nonisothermal conditions. The predicted concentration versus time profiles were in excellent agreement with the experimental results using model parameters that were independently determined from experiments or correlations.