Estimating PFOS Desorption Coefficients in Unsaturated Soils

The estimation of per‐ and polyfluoroalkyl substance (PFAS) linear soil–water partition coefficients ( K d values) is a key step in quantifying PFAS leaching from unsaturated soils. However, approaches for determining desorption K d values for PFAS have received little attention. Herein we focus on perfluorooctane sulfonate (PFOS) and consider the conventional approach of using the organic carbon‐normalized desorption coefficient ( K OC ) multiplied by the fraction of soil organic carbon ( f OC ) for estimating K d . Results based on soil desorption data from several studies showed that K OC values for PFOS spanned nearly two orders of magnitude, with a mean of 1600 L/kg and a standard deviation of 1600 L/kg. These K OC values showed no significant trend with soil f OC values for the range of f OC values (0.0007 to 0.047) in the compiled studies. Desorption equilibration times in excess of 24 h were observed in multiple studies, for which a 24‐h incubation period would have resulted in substantial (up to a factor of two) overestimation of PFOS desorption K OC values. Finally, comparison of K OC values derived based on a single desorption point (i.e., K OC is based on a single liquid: solid ratio) were up to an order‐of‐magnitude greater than K OC values based on the linear slope of a desorption isotherm (i.e., multiple desorption points from use of multiple liquid: solid ratios). This was due to hysteretic desorption behavior, where a fraction of the PFOS was not readily desorbed and the desorption isotherms exhibited positive nonzero intercepts. Such hysteresis indicates that adsorption K OC or K d values may be a poor indicator for desorption behavior for some soils. Together, these results suggest that caution should be used when relying on simple K OC ‐based approaches for estimating the K d value for PFOS, and single‐point desorption/leaching tests should consider the potential for slow (more than 24 h) and hysteretic desorption processes.