Machine learning maps nearly 10,000 reactions to improve CO₂-to-fuel conversion
In India, Bioengineer reports researchers at the Indian Institute of Science (IISc) have combined machine learning and quantum-mechanical simulations to map 9,389 chemical reactions involved in converting CO₂ and hydrogen into fuels and chemicals over a copper catalyst. The expanded reaction network improved model predictions, increasing estimated CO₂ conversion by around 40-fold compared with a model based on just 152 reactions, while correctly identifying methanol and carbon monoxide as major products. The research also uncovered a potentially important pathway in which molecular hydrogen transfers directly to reaction intermediates. The findings could guide the design of more effective catalysts for producing methanol and other fuels from CO₂.
Category: Research
