Prediction of suitable catalysts for the OCM reaction by combining an evolutionary approach and machine learning

Abstract

Catalytic systems are multidimensional and still difficult to interpret even by accomplished chemists. For years high throughput experimentation has been used to find new catalysts. We describe a method to use the concept of directed evolution to synthesize new catalysts for the oxidative coupling of methane in silico via a classical genetic algorithm. The evaluation of the novel catalysts is based on predicting the C2 yield with the help of a random forest algorithm.