Prediction of fluorophilicity of organic and transition metal compounds using molecular surface areas

A model for the prediction of fluorophilicity based on the regression of experimental values for 93 organic molecules against molecular surface area descriptors has been constructed. The model is predictively accurate, with R-2 = 0.94 and RMS = 0.638. A further model for the prediction of fluorophilicity of transition metal complexes was developed using the same methodology. Statistical analysis of this model showed it to be capable of predicting fluorophilicity with an error of ca. 0.6 In units. Finally, combination of these two models allows simultaneous prediction for organic and transition metal complexes. The accuracy of this model is comparable to previously published studies, but the ability to predict fluorophilicity for transition metal complexes represents a significant advance. In order to generate accurate surface area descriptors for fluorinated transition metal complexes, a QM/MM combination of B3LYP/Lanl2DZ for the metal centre with AMBER for fluorous chains was used to generate 3D structures. This method of geometry optimisation was seen to be capable of producing reliable structures with geometries similar to structures obtained front X-ray crystallography. The models so developed quantify the effects of size, hydrogen bonding, and exposed fluorine surface area on fluorophilicity, yielding a physically realistic model in agreement with previously published studies. (C) 2004 Elsevier B.V. All rights reserved.