Limitations of Current Approaches for Predicting Groundwater Vulnerability from PFAS Contamination in the Vadose Zone

Published literature for reported sorption coefficients (K-d) of eight anionic per- and polyfluoroalkyl substances (PFAS) in soil was reviewed. K-d values spanned three to five log units indicating that no single value would be appropriate for use in estimating PFAS impacts to groundwater using existing soil-water partition equations. Regression analysis was used to determine if the soil or solution parameters might be used to predict K-d values. None of the 15 experimental parameters collected could individually explain variability in reported K-d values. Significant associations between K-d and soil calcium and sodium content were found for many of the selected PFAS, suggesting that soil cation content may be critical to PFAS sorption, as previously noted in sources like Higgins and Luthy (2006), while organic carbon content was significant only at elevated levels (>5%). Unexplained discrepancies between the results from studies where PFAS were introduced to soil and desorbed in the laboratory and those that used material from PFAS-impacted sites suggest that laboratory experiments may be overlooking some aspects critical to PFAS sorption. Future studies would benefit from the development and use of standardized analytical methods to improve data quality and the establishment of soil parameters appropriate for collection to produce more complete data sets for predictive analysis.

Automated summary

The review of literature on reported sorption coefficients of eight anionic PFAS in soil revealed significant associations between soil cation content (calcium and sodium) and PFAS sorption, while organic carbon content was significant only at elevated levels. Discrepancies between laboratory and field studies suggest that critical aspects of PFAS sorption may be overlooked in laboratory experiments. Future studies would benefit from standardized analytical methods and appropriate soil parameters for more accurate predictions.