APPLICATION OF DFT AND MACHINE LEARNING TO PREDICT OPTIMUM OPERATING CONDITIONS FOR METHANE PYROLYSIS USING MOLTEN METALS FOR CARBON-FREE HYDROGEN PRODUCTION

The production of hydrogen from various feedstock involves the catalyzation of various reactions for the generation of the desired product. Experimental studies provide an understanding of the reaction mechanism and the nature of the reactants and products of the reaction. Computational studies involving DFT provide even greater insights into these reactions. The pyrolysis of methane over various catalysts requires an efficient means of screening catalysts for the reaction. The reaction is an endothermic reaction that is associated with a high demand for energy to drive the reaction. A process associated with a reduced demand for energy and increased yield of hydrogen will make the pyrolysis reaction more efficient and more industrially applicable. A combination of DFT and ML provides a means of establishing catalytic descriptor-based predictions for the reaction. The prediction of the relaxation energies of doped group 10, 11 and 12 transition elements from their initial structures with the adsorbate ions at the different stages of the reactions provides a predictable path for the calculation of the Turnover Frequency of the reactions, suggesting the preferred catalyst for the reaction. This study seeks to establish a model for predicting catalytic activity in methane pyrolysis reactions at various operational conditions.