Feature Analysis in High-Dimensional Data: Structure-Activity Relationships of Lewis Acid-Transition-Metal Complex-Catalyzed H2 Activation

Lewis acid-transition metal (LA-TM) catalystshavebeen proven to have an advantage in catalyzing hydrogen activation.Herein, a high-dimensional structure-activity relationshipstudy is performed for LA-TM-catalyzed hydrogen activationby density functional theory calculations. The DPB-Ni complexis taken as the representative catalyst, and the explored Lewis acidsites and transition-metal centers include B, Al, Ga and Ni, Pd, Pt,respectively. Totally, four general hydrogen activation mechanismsare systematically studied among the nine catalytic systems. The Ga-Nisystem undergoes the lowest free energy of activation (11.0 kcal/mol),which is considered to be the optimal combination of the Lewis acidsite and transition-metal center. Furthermore, more than 100 parametersare used to analyze the structure-activity relationship, includingthe physical structure, the bond order, the atom charge, and manyother properties. Key parameters of important structures are dug outto show a high correlation with the activity of the LA-TM systems,including the M-H-2 distance, the H-H bondlength, the second-order perturbation stabilization energy of M-H-2, the bond order of the LA-TM, and so on. The multivariableanalysis indicates that the feature related to the basic elementalproperties and the global feature codetermine the activity of thecatalyst. In the LA-TM system, the combination of Ip (LA)/Ip (TM) (Ip, the first ionization energy, the feature related to basic elementalproperties) and the chemical hardness (the global feature) can betterexplain the activity of the catalyst. The Ip (LA)/Ip (TM) reflects the difficultyof breaking the LA-TM bond, affecting the reaction site ofactivating hydrogen. The hardness reflects the stability and reactivityof LA-TM-RC complexes. The above two featureswith the addition of the LA-TM bond length (the local feature)can better reflect the activity of the LA-TM system-catalyzedH(2) activation. The feature combinations and the methodof multidimensional data analysis should be informative guidance forthe rational design of efficient LA-TM catalysts for H-2 activation.