Performance of chromatographic systems to model soil–water sorption

A systematic approach for evaluating the goodness of chromatographic systems to model the sorption of neutral organic compounds by soil from water is presented in this work. It is based on the examination of the three sources of error that determine the overall variance obtained when soil–water partition coefficients are correlated against chromatographic retention factors: the variance of the soil–water sorption data, the variance of the chromatographic data, and the variance attributed to the dissimilarity between the two systems. These contributions of variance are easily predicted through the characterization of the systems by the solvation parameter model. According to this method, several chromatographic systems besides the reference octanol–water partition system have been selected to test their performance in the emulation of soil–water sorption. The results from the experimental correlations agree with the predicted variances. The high-performance liquid chromatography system based on an immobilized artificial membrane and the micellar electrokinetic chromatography systems of sodium dodecylsulfate and sodium taurocholate provide the most precise correlation models. They have shown to predict well soil–water sorption coefficients of several tested herbicides. Octanol–water partitions and high-performance liquid chromatography measurements using C18 columns are less suited for the estimation of soil–water partition coefficients.