A data-driven high-throughput workflow applied to promoted In-oxide catalysts for CO hydrogenation to methanol

We propose a novel high-throughput workflow, combining DFT-derived atomic scale interaction parameters with experimental data to identify key performance-related descriptors in a CO 2 to methanol reaction, for In-based catalysts. Utilizing advanced machine learning algorithms suitable for small datasets, secondary descriptors with high predictive power for catalytic activity were constructed. These descriptors, which highlight the crucial role of hydroxyl sites, can be applied to designing new materials and to bringing them to the test with high-throughput screening, paving the path for accelerated catalyst design. To facilitate accelerated catalyst design, a combined computation and experimental workflow based on machine learning algorithms is proposed, which detects key performance-related descriptors in a CO 2 to methanol reaction, for In 2 O 3 -based catalysts.