Prediction of nematic transition temperatures in thermotropic liquid crystals by a heuristic method

A quantitative structure-property relationship (QSPR) study was performed on 42 homogeneous thermotropic liquid crystals (TLCs) to predict their nematic transition temperatures (T N) using molecular descriptors calculated by CODESSA software. A heuristic method (HM) was applied to select descriptors and generate models. The selected five descriptors reflect the constitutional, steric and electronic characteristics that affect the nematic transition phenomena. The high correlation coefficient, R 2, of 0.9881 for the training set indicates that a linear relationship exists between the structural information and T N in TLCs. Satisfactory results were obtained. For the test set, this model gave an R 2 of 0.96, a root mean squared error of 6.3654 and an absolute average relative deviation of 9.2017%. The results provide a simple, practical and effective method for the prediction of T N of TLCs and the approach can be extended to other QSPR investigations.