Evaluation of extraction workflows for quantitative analysis of per- and polyfluoroalkyl substances: A case study using soil adjacent to a landfill.

Specific aspects of previously reported extraction workflows, for measurement of per- and polyfluoroalkyl substances (PFAS) in solid matrices, have not been adequately interrogated. The objective of this study was to explore the importance of each workflow step in providing the most appropriate extraction for a comprehensive set of PFAS (51 different species) in soil. We compared different procedures, including two pre-extraction set ups (overnight handling of samples prior to extraction), two extraction solvents (methanol (MeOH), and acetonitrile (ACN)), two extraction solvent volumes (10 mL and 8.5 mL), and two post-extraction cleanup strategies (ENVI-Carb and ion-pair). Of the 51 species targeted, 21 were at quantifiable levels in soil samples collected adjacent to a landfill, of which 13 PFAS were consistently detected among the different extraction workflows. Overall, results showed no significant difference in PFAS concentration between different extraction solvents and cleanup strategies. Perfluoropentanoic acid, perfluorohexanoic acid, and perfluorooctanoic acid had the highest concentrations in all extraction workflows, accounting for nearly 13%, 38%, and 17% of the total monitored PFAS (ΣPFAS), respectively. While final concentration values were similar across methods, recovery and accuracy studies showed that MeOH had the best recovery, with 88% of the isotopically labeled PFAS standards showing extraction recovery within the acceptable range of 80% to 120% (compared to 14% of isotopically labeled PFAS standards in workflows using ACN). Upon examination of additional cleanup steps, 67% of monitored PFAS (out of 51 total PFAS monitored), on average, exhibited higher accuracy (relative error ≤20%) using ENVI-Carb clean up (in comparison with 51% in workflows using ion pair clean up). Results also demonstrated that larger volumes of MeOH (and subsequent re-extractions) did not yield a better recovery, enabling a reduction in overall analysis time and cost in comparison to many published methods.