Batch equilibrium experiments and modeling reveal weak temperature dependence of cyclic volatile methylsiloxane sorption in soil/sediment organic carbon-water systems.

The organic carbon normalized partition coefficient, K, describes the equilibrium distribution of a chemical between water and organic carbon in soil or sediment. It is a key parameter in evaluating chemical persistence, mass distribution, and transport using multimedia fate and transport models. Considerable uncertainty remains about the K values of cyclic volatile methylsiloxane (cVMS) compounds, and in particular the dependence of K on temperature. In this study, we used a batch equilibrium (BE) method to measure K values and their temperature dependence between ∼5 and 25 °C for octamethylcyclotetrasiloxane (D4) and decamethylcyclopentasiloxane (D5) with soil and sediments. Approximate log K values at 25 °C were 4.5-5.0 for D4 and 5.5-6.1 for D5 with different sorbents, and decreased by 0.3 log units or less at 4-5 °C. Enthalpies of sorption, ΔH, obtained for the different sorbents ranged from +7.2 to +16 kJ mol, with average values of +7.9 and +13 kJ mol for D4 and D5, respectively. These values differ in magnitude and direction from those reported elsewhere based on K values determined by a novel dynamic purge-and-trap (PnT) method, but are consistent with predictions based on their solvation properties. A new fugacity-based multimedia model incorporating sorption/desorption kinetics was developed and used to predict concentrations in the phases of BE and PnT systems during desorption of cVMS under different experimental and ideal conditions. Model simulations suggested that K values for cVMS compounds derived from the PnT systems could be influenced by sorption disequilibrium between water and solids controlled by desorption rates from the particle phase to water, and subsequent losses due to volatilization and degradation. This has the potential to result in overestimation of K values when fitting the experimental data of cVMS mass remaining in a PnT system over time, which could explain the observed differences between the methods.