Artificial neural network-aided development of supported Co catalyst for preferential oxidation of CO in excess hydrogen

Preferential oxidation (PROX) of 1 vol% CO was investigated using the stoichiometric amount of O2 in excess hydrogen. Cobalt supported on alkali earth metal carbonate, especially Co/SrCO3, showed high selectivity to PROX of CO while the assistance of methanation of CO was necessary to achieve over 99% conversion of CO. Cobalt loading and preparation conditions of Co/SrCO3 was optimized by using a full factorial design of experiment, an artificial neural network and a grid search. Over 99% O2 conversion and 78% selectivity for PROX at 200 °C was achieved by 3.2 mol% Co/SrCO3 pretreated at 345 °C.