Machine learning in heterogeneous catalysis
Literature review of scientific articles on heterogeneous catalysis using machine learning or high-throughput synthesis and characterization.
Developing machine learning for heterogeneous catalysis with experimental and computational data
Link to Nature Reviews Chemistry and ChemRxiv
The review systematically compares published studies in terms of machine learning methods, input and output features, dataset size, reaction type, and reported accuracy. We identified 41 scientific articles on heterogeneous catalysis that report unitless R2 values for machine learning models, allowing for an objective comparison. The data compiled from these papers is displayed in the interactive plots below and in Figure 3 of the review. The label CO and CO2 reactivity includes any catalysis with CO2 and CO as substrates, like syngas reactions, except where methane is involved; Methane reactivity includes all the studies where methane is either reactant or product; Water splitting includes hydrogen and oxygen evolution reactions and oxygen reduction reaction.
X-axis: Y-axis:
Download the data:
Upload your own data, keeping the same labels and format: