Accelerated Oxide-Zeolite Catalyst Design for Syngas Conversion by Reaction Phase Diagram Analysis and Machine Learning

Oxide-zeolite (OXZEO) catalyst design concept provides an alternative approach for the direct syngas-to-olefins (STO) with superior selectivity. Enhancing the activity of oxide components remains a critical and long-pursued target in this field. However, rational design strategies for optimizing oxides and improving the catalyst performance in such complex reaction networks are still lacking. We employed energetic descriptors such as the adsorption energies of CO* and O* (G_adCO* and G_adO*) through reaction phase diagram (RPD) analysis to predict the catalyst performance. The prediction was initially validated by the catalytic activity trends measured by experiments. Machine learning (ML) was further utilized to accelerate the screening of new catalysts. Ultimately, Bi-doped and Sb-doped ZnCrO_x were theoretically predicted as optimized oxide candidates for the OXZEO reaction, which was experimentally verified to be more active than the currently best ZnCrO_x counterpart. This work demonstrated enhanced OXZEO catalysts for STO as well as a research paradigm integrating theory and experiment to optimize bifunctional catalysts for complex reaction networks.