Improvement of aquatic pollutant partition coefficient correlations using 1D molecular descriptors — chlorobenzene case study

Abstract

AbstractPartition coefficients between environmental compartments are essential parameters in any predictive models on pollutants’ fate in various emission scenarios. When sufficient experimental data are not available, empirical algebraic models are capable of predicting the pollutant partitioning characteristics based on bulk physico-chemical properties or various molecular structural features. When the use of sophisticated rules based on detailed 2D–3D molecular descriptors is not available as a quick option, inexpensive, simple correlations based solely on octanol-1-ol (octanol)-water partition coefficients (K ow) are extensively employed. The present study investigates enhancing the adequacy of such hydrophobicity-based models by adding simple 1D descriptors, readily identifiable by inspecting the substance structure (i.e. the number of chlorine atoms bound to aromatic rings, or the number of aromatic 5- or 6-atom rings, etc.), in addition to the pollutant’s solubility in water. Exemplification is made for predicting the water-biota (fish)-sediment partition coefficients for chlorobenzenes (CBz).