Theoretical study of CO2 hydrogenation to methanol on modified Au/In2O3 catalysts: Effects of hydrogen spillover and activation energy prediction for hydrogen transfer

With the increasing attention in environmental issues caused by CO₂ emissions, methanol conversion by CO₂ hydrogenation is an effective strategy to solve this existing energy dilemma. The rationale behind hydrogen spillover on methanol synthesis is unraveled via density functional theory (DFT) calculations in this work, furthermore, the activation energy of hydrogen transfer process as affected by spillover is also summarized in a general paradigm for facilitating the understanding of hydrogenation characteristics. The results demonstrate that the spillover strategy significantly facilitates the hydrogenation reaction by supplying available hydrogen adatoms. This effect is particularly pronounced during the stage when OH is formed directly at the substrate site and combines with H to produce H₂O, leading to a substantial reduction in activation energy from the initial 3.74 eV to 0.78 eV. In addition, a comprehensive predictive model for the kinetic characteristics of hydrogen spillover process is established based on the machine learning algorithm and SISSO guidance. By employing the combined approach of SISSO and neural network, we have achieved a stable prediction performance for activation energy with R² = 0.99 and RMSE = 0.07 eV. The variable of $Ch{g}_{FS}^{Au}$ is identified as the most representative factor in describing the activation energy, demonstrating a correlation coefficient of -0.60. The extended multidimensional expression of Dist_Au further highlights its close connection to activation energy, achieving an RMSE value of 0.41 eV. To sum up, this work elucidates the possible thoughts of catalyst design with spillover effect and gives reference for the description screening towards the chemical reactions similar to hydrogen spillover.