QSPR based on support vector machines to predict the glass transition temperature of compounds used in manufacturing OLEDs

Abstract

Electroluminescent polymeric materials have many applications, especially in organic light-emitting diodes (OLEDs). The use of these materials for this purpose requires knowledge of properties such as the thermal stability of materials, which can be studied through the glass transition temperature (T g). A quantitative structure–property relationship (QSPR) model based on the support vector machine (SVM) was developed to predict the T g of polymeric materials used in the manufacture of electroluminescent devices (OLED). A total of 66 structures were selected from the literature. The structures were drawn in ChemSketch; their three-dimensional coordinates were calculated using E-CORINA and molecular descriptors were calculated using E-Dragon. The selection of the best descriptors was performed using the BestFirst method and prediction models were developed using the SVM in WEKA software. The best model was obtained using the following SVM parameters: C = 1.0, γ = 0.06 and ϵ = 0.001.The obtained model showed a correlation coefficient (R 2) of 0.9626, a root mean square error of 0.0114, a mean absolute error of 0.0156 and a predictive squared correlation coefficient (Q 2) of 0.909.These results show that QSPR using the SVM is a powerful tool that can be used to predict the T g of materials used in OLED fabrication.