Refractive indices of diverse data set of polymers: A computational QSPR based study

In silico prediction of physico-chemical characteristics of materials has become an important tool in modern chemist's arsenal to design functional materials with desired properties. Here a QSPR model was developed for prediction of refractive indices (n) in a diverse set of organic polymers. Theoretical descriptors used in the model were generated from the structure of the repeating units of the polymers. A four variable model was developed for prediction of refractive indices with R-2 = 0.932 and (Q(LOO)(2) = 0.914. The predictive ability and robustness of the proposed model was validated using various methods, including cross-validation and y-randomization experiments. The R-ext(2) for the test set was found to be 0.882, which confirms the statistical significance of the model. For refractive index, the polarizability and sp(2) hybridized carbon atoms were the most influential properties. In addition, four small virtual libraries of the novel polymers were designed and their n values were predicted using the developed model. (C) 2017 Elsevier B.V. All rights reserved.