In Silico Models for Prediction of Methanol Yield in CO₂ Hydrogenation Reaction Using Cu-Based Catalysts

CO₂ hydrogenation to methanol is instrumental in mitigating carbon emissions and providing a renewable source of clean fuel, methanol. Though Cu-based catalysts proved to be economical and efficient catalysts for this reaction, it has the disadvantage of low catalyst efficiency and sintering. In this study, we developed different six machine learning (ML) models for the prediction of methanol yield (%) from CO₂ hydrogenation for Cu-based catalysts. The gradient boost random trees model outperformed other ML models with accuracy R² and RMSE of 0.96, 0.71 on train data and 0.75, 1.75 on test data. Pressure, metal:support ratio, active metal composition, GHSV and reaction temperature were found to be influential parameters for optimization of methanol yield. The prediction capability of this model is also validated based on unseen experimental data with varied input parameters and the predictions are good enough with R² and RMSE of 0.9 and 1.14. Therefore, this model can be regarded as a valuable solution to guide experimental design without actual experimentation for Cu-based catalysts.