Regresar

Molecular and Descriptor Spaces for Predicting Initial Rate of Catalytic Homogeneous Quinoline Hydrogenation with Ru, Rh, Os, and Ir Catalysts

Abstract:

Developing highly active catalysts for quinoline hydrogenation is crucial for efficient hydrogen carrier technologies and clean fossil fuel hydrodenitrogenation. In this work, we employed Tensor Algebra-based 3D-Geometrical Molecular Descriptors (QuBiLS-MIDAS) to develop Quantitative Structure-Property Relationship (QSPR) models predicting the initial rate of homogeneous quinoline hydrogenation catalyzed by transition metal complexes of Ru, Rh, Os, and Ir. A data set of 32 catalytic precursors was used: 25 for model training (training set) and 7 for external validation (testing set). Multiple linear regression analysis yielded a model with good predictive ability for the training set (R2 = 0.90) and satisfactory external validation for the testing set (Q<inf>EXT</inf>2 = 0.86). The model’s descriptors highlighted the importance of hardness, softness, electrophilicity, and mass in predicting catalytic activity. The virtual screening revealed that Rh and Ir complexes with π-acidic ligands (e.g., olefins, diolefins, and η5-Cp) and nitrile ligands exhibited the highest predicted catalytic activity, suggesting potential for further improvement through ligand structural modification. Notably, iridium complexes, particularly those with tri(cyclopropyl)phosphine ligands, demonstrated significant potential for hydrogen storage, transport, and production, underscoring their relevance in sustainable energy systems. These findings demonstrate the potential of the QuBiLS-MIDAS approach for in silico design of efficient catalysts for quinoline hydrogenation processes.