Quantitative structure–property relationship models to predict thermodynamic properties of some mono and polycyclic aromatic hydrocarbons using genetic algorithm‐multiple linear regression

Abstract

AbstractPolycyclic aromatic hydrocarbons (PAHs) are a class of diverse organic compounds. Quantitative structure–property relationship (QSPR) models are used for modeling and predicting the thermodynamic properties of 45 monocyclic aromatic hydrocarbons (MAHs) and PAHs. Thermodynamic properties of PAHs such as heat capacity (C v; J mol−1 K−1), thermal energy (Eth; kJ mol−1), and entropy (S; J mol−1 K−1) are calculated at the HF level of theory and 6‐311G *basis sets using the Gaussian 09 program. A set of molecular descriptors is calculated for selected compounds using the software Dragon. The genetic algorithm (GA) and backward stepwise multiple linear regression (MLR) techniques are used to obtain suitable QSPR models. The predictive ability of the models is checked using several criteria for internal and external model validation. The results show that three descriptors (Uindex, nAT, HTe) can be efficiently used for estimating S, one descriptor (HTe) for modeling Eth, and also one descriptor (Uindex) for predicting C v of the considered compounds. The results and discussion lead to the conclusion that the models established by GA‐MLR have good correlation of the thermodynamic properties, which means QSPR models can be efficiently used for predicting the above‐mentioned properties and designing new molecules of MAHs and PAHs.