Development and evaluation of MTLSER and QSAR models for predicting polyethylene-water partition coefficients

Current study was aimed to make further improvements in measuring low density polyethylene (LDPE) -water partition coefficient (K_PE-w) for organic chemicals. Modified theoretical linear solvation energy relationship (MTLSER) model and quantitative structure activity relationship (QSAR) model were developed for predicting K_PE-w values from chemical descriptors. With the MTLSER model, α (average molecular polarizability), μ (dipole moment) and q^- (net charge of the most negative atoms) as significant variables were screened. With the QSAR model, main control factors of K_PE-w values, such as CrippenLogP (Crippen octanol-water partition coefficient), CIC0 (neighborhood symmetry of 0-order) and GATS2p (Geary autocorrelation-lag2/weighted by polarizabilities) were studied. As per our best knowledge, this is the first attempt to predict polymer-water partition coefficient using the MTLSER model. Statistical parameters, correlation coefficient (R²) and cross-validation coefficients (Q²) were ranging from 0.811 to 0.951 and 0.761 to 0.949, respectively, which indicated that the models appropriately fit the results, and also showed robustness and predictive capacity. Mechanism interpretation suggested that the main factors governing the partition process between LDPE and water were the molecular polarizability and hydrophobicity. The results of this study provide an excellent tool for predicting log K_PE-w values of most common hydrophobic organic compounds, within the applicability domains to reduce experimental cost and time for innovation.