Linear free energy relationships used to evaluate equilibrium partitioning of organic compounds

In environmental chemistry, one often wants to interpret or predict equilibrium partitioning of organic compounds between any two phases. Hence, one needs to understand the partition variability that stems from using different types of compounds and the variability that arises from looking at different natural phases, e.g. different soil organic matter. It is current practice in environmental chemistry to evaluate equilibrium partitioning with the help of double logarithmic correlations between the unknown partition constant and a well-known partition constant of the compounds, e.g., partitioning between natural organic matter and water or air is correlated with the octanol/water or octanol/air partition constant, respectively. However, these relationships (in the following called one-parameter LFERs) can only predict the compound variability within a single substance class. They supply no means to understand the variability between substance classes or the variability between different natural organic phases. The reasons for these limitations are that (a) the complete compound variability cannot be described by a single parameter because partitioning results from different kinds of interactions that vary independently from each other and (b) the specific properties of the studied phase are represented in the slope and intercept of the double logarithmic correlation and not in a variable parameter. In contrast to one-parameter LFERs, polyparameter LFERs are based on a concept that considers all interactions involved in partitioning by separate parameters. They allow for predicting the complete compound variability by a single equation, and they also provide the possibility to evaluate and predict the variability in the sorption characteristics of different natural phases. Thus future research in the field of environmental partition processes should focus on adapting and improving the more comprehensive polyparameter LFERs rather than trying to refine existing one-parameter LFERs.