Researchers at the Indian Institute of Science have mapped thousands of chemical reactions involved in turning CO2 into fuel and chemicals on a copper catalyst. The team built a computational framework covering 9,389 elementary reaction steps, then used machine learning to estimate energy barriers that would be too expensive to calculate with quantum mechanics alone. With the larger network, the model predicted methanol and carbon monoxide as the main products, which matches experimental results better than the smaller 152-reaction model did. The work also surfaced an unusual hydrogen transfer pathway that only appears when the full reaction landscape is included. The authors say the same approach could be adapted to other catalytic processes, including CO2 reduction on different catalysts, nitrogen reduction, and water splitting. For anyone following CO2 utilization and low-carbon fuel research, this is a useful step toward catalyst design that is less guesswork and more systematic.
