Hydrogen sulfide is a significant byproduct of oil and gas production and is typically recovered as elemental sulfur, a low-value commodity. In recent years, there have been efforts to upgrade H2S through elemental decomposition to S and H2, an essential energy carrier in a sustainable economy. Among the promising approaches is thermocatalytic looping, which involves a sulfide-based redox pair. Unfortunately, the search for sulfides capable of facilitating this conversion is progressing slowly, and primarily focusing on monometallic sulfides. With a few notable exceptions, the field of bimetallic sulfide remains largely unexplored. In this study, a machine learning framework is employed to explore the material space of mono and bimetallic sulfides. The workflow begins by mining sulfides from the Materials Project database, allowing the workflow to be benchmarked using formation enthalpies derived from established density functional theory calculations. Through the machine learning framework, the number of bimetallic sulfide redox pairs considered is expanded from 102 cases in the Materials Project database to 105 cases. This expansion allows for the identification trends that can serve as guidelines for future research and helps prioritize materials for experimental testing.
Keywords: chemical looping; high‐throughput materials screening; hydrogen sulfide decomposition; machine learning; redox catalysis.
© 2024 Wiley‐VCH GmbH.