Machine learning prediction and experimental verification of Pt-modified nitride catalysts for ethanol reforming with reduced precious metal loading

Abstract

Ethanol is the smallest molecule containing C–O, C–C, C–H, and O–H bonds present in biomass-derived oxygenates. The development of inexpensive and selective catalysts for ethanol reforming is important towards the renewable generation of hydrogen from biomass. Transition metal nitrides (TMN) are interesting catalyst support materials that can effectively reduce precious metal loading for the catalysis of ethanol and other oxygenates. Herein theoretical and experimental methods were used to probe platinum-modified molybdenum nitride (Pt/Mo2N) surfaces for ethanol reforming. Computations using density-functional theory and machine learning predicted monolayer Pt/Mo2N to be highly active and selective for ethanol reforming. Temperature-programmed desorption (TPD) experiments verified that ethanol primarily underwent decomposition on Mo2N, and the reaction pathway shifted to reforming on Pt/Mo2N surfaces. High-resolution electron energy loss spectroscopy (HREELS) results further indicated that while Mo2N decomposed the ethoxy intermediate by cleaving C–C, C–O, and C–H bonds, Pt-modification preserved the C–O bond, resulting in ethanol reforming.