Sorption Equilibrium of 190 Organic Vapors in Soil Organic Matter as a Function of Temperature and Humidity: Experiments and Modelling.
Christian Niederer, Kai-Uwe Goss, and René P. Schwarzenbach. Swiss Federal Institute of Technology ETH, Universitätsstrasse 16, Zurich, 8092, Switzerland
Sorption into soil organic matter is an important process for the fate of non-ionic, organic pollutants in the environment. The extent of the sorption is determined by van-der-Waals and H-bond (electron-donor/acceptor) interactions between the pollutants and the soil organic matter. In this study (Leonardite Humic Acid)/(air) partition coefficients of more than 190 organic compounds with a wide range of functional groups were determined using an inverse gas chromatography (IGC) method at temperatures between 5°C and 75°C. By varying the relative humidity in the IGC system, the influence of the hydration state of the humic acid on its sorption properties was investigated. To the best of our knowledge this is by far the largest and most diverse sorption data set for soil organic matter that has ever been measured with one consistent method. Many substances of direct environmental concern such as phthalates, pesticides, polychlorinated benzenes, polychlorinated phenols or nitroaromatic compounds are included in the dataset. The humic acid/air partition coefficients range over seven orders of magnitude. Polar compounds generally sorbed stronger than non-polar compounds due to specific H-bond (electron-donor/acceptor) interactions with the humic acid. The relative humidity had a rather small influence (less than a factor of three) on the experimental results. No glass transition was observed in Leonardite humic acid in the temperature range from 5 to 75°C. A comparison of our experimental data with classical predicting tools for sorption in soil organic matter based on octanol/air partitioning or molecular connectivity (EPI SuiteTM) showed that these models cannot describe our measured data satisfactorily. In contrast, our experimental data can successfully be described by a polyparameter Linear-Solvation-Energy Relationship (pp-LSER) that identifies the contributions from van-der-Waals and electron donor/acceptor interactions. The pp-LSER approach also enabled us to estimate the temperature dependency of the sorption process. However, this polyparameter approach has still two shortcomings: a) for every sorbate of interest the mentioned van-der-Waals and electron donor/acceptor descriptors have to be determined in separate experiments if they are not tabulated in the literature; b) the required sorbent descriptors can only be determined from a set of experimental sorption coefficients measured for at least 40 different reference compounds. Therefore, a tool that is able to predict sorption coefficients independent of any experimental calibration data would be highly desirable. Principally, such a tool is provided by the quantum-chemical software COSMOtherm that considers the electrostatic interactions of the solvent with the solute based on interacting molecular surface charges. For calculations of sorption coefficients COSMOtherm demands 3D-input structures for the sorbent as well as the sorbate molecules. Here, we report for the first time an evaluation of the performance of COSMOtherm in predicting sorption coefficients for soil organic matter. For this purpose we derived a model molecular structure of a Leonardite humic acid monomer based on published spectrometric and elemental analysis data (1H-NMR,13C-NMR, IR, UV/VIS, acidic function analysis). A suitable model monomer should represent all the major physical-chemical properties of Leonardite HA (e.g., electron donor/acceptor properties). The monomer size had to be kept in the range of 35 carbon atoms due to the exponential increase in computation with increasing molecular size. A comparison of the calculated sorption coefficients with our experimental data show that despite of slight systematic deviations COSMOtherm can predict the experimental data within a factor of 4-5. In addition, the relative precision of the calculated partition coefficients is excellent.