Diffusion of Sorbing Organic Chemicals in the Liquid and Gaseous Phases of Repacked Soil

Transport models for sorbing organic chemicals in soil require accurate predictions of the diffusion and sorption processes in both the liquid and gaseous phases. In this study, the ability of recently-presented diffusivity models in combination with equilibrium sorption models to predict the effective (i.e., including sorption effects) diffusion coefficient, D ₑff, as a function of soil-water content, θ, is tested for different sorbing organic chemicals in different soils. The water-induced linear reduction (WLR) gas diffusivity model, combined with a two-component (hydrophobic and vapor) equilibrium sorption model, well described short-term (<24 h) D ₑff(θ) of two volatile organic compounds, toluene and trichloroethylene (TCE), in Yolo loam (fine-silty, thermic Typic Xerorthent). Further extended with the constant slope impedance factor (CSIF) solute diffusivity model, the resulting two-phase diffusion and two-component equilibrium sorption (DATES) model accurately predicted short-term (48 h) D ₑff(θ) of a semivolatile pesticide, lindane (C₆H₆Cl₆), in Gila silt loam (coarse-loamy, thermic Typic Torrifluvents). To test the DATES model for longer-term data, D ₑff(θ) of naphthalene (C₁₀H₈) in Lerbjerg sandy clay (L5) was measured at different soil-water contents and incubation times in air-tight diffusion cells. Because of sorption kinetics, both the apparent hydrophobic sorption coefficient, K D, derived from the column diffusion experiments and K D derived from batch desorption experiments increased with naphthalene-soil contact time for t > 700 h. Consequently, an increasing K D value with time was required in the DATES model to obtain successful D ₑff(θ) predictions. Since chemical-soil contact time and adsorption-desorption history typically vary in each compartment of a soil profile, transport models, including sorption kinetics, will become very complex and DATES with a time-dependent K D relation may represent a useful model alternative.