Gas phase heterogeneous catalyst performance testing in laboratory fixed-bed reactors

Activity, selectivity and stability are invariably among the key factors of the performance of a catalyst. In the development of catalysts these properties are often screened for a range of materials and formulations. Interpretation of these key performance indicators are prone to various confounding effects. Here, performance testing of solid, porous catalysts for gas phase reactions in tubular fixed bed reactors is considered. Transport limitations and particularly internal mass transfer limitations are often cited in this case. Many have given general discussions and guides for effective catalyst performance testing, reviewed or put forward theoretical descriptions for transport phenomena and have measured and correlated associated transport coefficients. Some quantitative requirements and the relative importance of different effects have been found to remain unclear. Here, some of these aspects are addressed by the development of 3 catalyst testing criteria. Specifically, an upper limit is derived for the chemical conversion in a firstorder reaction such that differential rate conditions are established, a lower limit on the chemical conversion is applied to limit the loss of precision in conversion measurements, and an expression is derived to limit the effect of pressure drop across a catalyst bed on the observed rate of a first-order reaction. The prevalence and sensitivity of these and other transport limitation criteria were investigated theoretically in the context of the low-temperature (LT) water-gas shift (WGS) reaction over a Cu/ZnO/Al2O3 catalyst in a laboratory scale performance test. Factorial combination of some commonly manipulated experimental parameters (reactant feed rate, temperature, catalyst particle size, catalyst loading, dilution fraction and reactor tube size) was employed in this regard. The upper conversion limit, the internal mass transfer criterion and the radial heat transfer criterion were found to be particularly severe. So too, to a lesser extent, were the axial dispersion and pressure drop criteria, and the lower conversion limit. The sensitivity analysis indicated optima in the varied experimental parameters and yielded insights into effective control of different effects by selection of process conditions. Application of the set of criteria in an experimental performance test was demonstrated using a proprietary medium-temperature (MT), WGS catalyst under reaction at temperatures of 275 °C, 300 °C and 375 °C, 1 atm total pressure, dry feed composition of 10% CO, 10% CO2, 70% H2, 10% N2, steam-to-dry gas ratio of 0.5 and 158 h-1 weight hourly space velocity (WHSV). The catalyst was found to have near total selectivity towards the WGS reaction with activities of 12.2 ± 1.1, 17.1 ± 0.5 and 24.9 ± 1.5 µmol/s.gcat at 275 °C, 300 °C and 375 °C respectively. This corresponds to an activation energy of 39 ± 2 kJ/mol; a value within range of what is reported in literature for similar catalysts. This experiment also served to compare experimental and predicted internal mass transfer limitations by testing catalyst particles of different mean sizes. This catalyst as well as a CuO/ZnO/Al2O3 catalyst precursor was characterised in respect of their pore size distributions (N2 physisorption and mercury intrusion porosimetry (MIP)), particle size distributions (by photo- and microscopic analysis), bulk and particle densities and product gas compositions (by gas chromatography) to enable evaluation of the various criteria employed. Evaluation of the various criteria indicated that, theoretically, the considered confounding effects had a negligible effect on the measured catalytic activities for the catalyst sample with the smallest mean particle size, while the larger particles experienced only internal mass transfer limitations. Different models considered for effective diffusivities all under-predicted values when compared to the effective diffusivities inferred from the reaction-diffusion experiments. Predictions ranged to within factors of 3 – 20 of the experimental values, depending on whether pore size distribution data were derived from MIP or physisorption data. Here, the lack of characterisation of the macro-porosity by N2 physisorption resulted in more severe under-estimations of the effective diffusivities than the equivalent estimations made with MIP data. The best prediction was made by the ‘parallel-path pore’ model by Johnson & Stewart (1965) using MIP data. Predictions of internal mass transfer limitations varied in a similar manner. It is noted that the simplifications of the highly complex porous catalyst by these model combinations introduce large sources of error in the prediction of internal mass transfer limitations