Efficient way of designing fullerene derivatives based on simplified DFT calculations and QSPR modeling

The influence of molecular structure of cyclopropene derivatives on solubility of C60 and C70 methanofullerenes is presented in the framework of quantitative structure–property relationship (QSPR) modeling. The best subset of molecular descriptors was selected with genetic algorithm (GA) from a variety of theoretical molecular descriptors, calculated for the chemical structures of cyclopropane derivatives. The QSPR model developed with multiple linear regression (MLR) was characterized by high robustness (Q2CV = 0.86) and the external predictivity (Q2EXT = 0.75). The external validated model has been applied to predict solubility (S) of fullerene derivatives for which experimentally determined values of S have been unavailable. Since the two descriptors utilized by the QSPR model are calculated for relatively small structural fragments, the application of this approach does not require extensive and costly quantum-mechanical calculations to be performed for the whole molecular structures of C60 and C70 fullerene derivatives. We believe that both the finally obtained nano-QSPR model and the suggestions considering the dependence of the solubility of fullerene derivatives in chlorobenzene on the molecular structures of the substituents will find practical applications in designing new materials.