Prediction of effective additives to a Ni/active carbon catalyst for vapor-phase carbonylation of methanol by an artificial neural network

Effective additives were investigated in order to suppress the formation of methane, a major byproduct of vapor-phase carbonylation of methanol with a Ni/active carbon (AC) catalyst. An artificial neural network was applied to relate the physicochemical properties of an element (X) with experimentally determined methane selectivity of the catalyst containing the element (Ni-X/AC). The trained artificial neural network succeeded in finding Sn as effective based on the training data where information about methane selectivity of Ni-Sn/AC was not included.