Abstract.
A Quantitative Structure–Property Relationship (QSPR) analysis and study of polycyclic aromatic hydrocarbons (PAHs) is presented. Three physicochemical properties related to their environmental impact are studied: boiling point (bp), octanol–water partition coefficient (log K_ow) and retention time index (RI) for reversed-phase liquid chromatography analysis. The geometry of all PAHs were optimized by the semi-empirical method AM1 and used to calculate thermodynamic, electronic, steric and topological descriptors: HOMO and LUMO energies and the GAP between them, molecular hardness, polarizability, atomic charges, connectivity index, volume and surface area among others. After variable selection, principal component regression (PCR) and partial least squares (PLS) with leave-one-out crossvalidation were used for building the regression models.
The regression coefficients obtained for the models were 0.995 (PCR and PLS) for bp, 0.975 (PCR) and 0.976 (PLS) for log K_ow, and 0.898 (PCR and PLS) for RI. Finally, the models were used to predict these properties for those compounds for which experimental measurements are still unknown.

Keywords.
Polycyclic Aromatic Hydrocarbons; Principal Component Regression; Partial Least Squares; Molecular Descriptors; AM1; WHIM-3D; Physicochemical Properties.

Keywords Plus.
Molecular Descriptors; Prediction; Toxicity; Polar; QSAR; Regression; Pollutants.