PFAS quantitation with diffusive gradients in thin-film passive samplers: Capturing time-weighted average
Brianna J Harris1, Samuel D Hodges1, David G Wahman2
1Department of Civil Engineering, University of Arkansas, Fayetteville, AR 72701, USA.
Abstract:
Diffusive gradients in thin-films (DGT) passive samplers are used to quantify analytes in water, but their capabilities for capturing time-weighted average concentrations (CDGT) of per- and polyfluoroalkyl substances (PFAS) at low ng L-1 levels are unknown. For 32 PFAS, DGT passive sampler gel layer diffusion coefficients (DGel) ± 95 % confidence intervals (CIs) were determined using two-compartment diffusion cell tests analyzed with a non-steady-state finite difference model (FDM), which was previously shown to produce DGel estimates with less error than traditional methods relying on a pseudo-steady-state flux assumption. For each PFAS, the FDM also determined the normalized weighted sum of squared errors (WSSE × n-1), a goodness of fit measure. Eleven PFAS had adequate FDM fits (WSSE × n-1 < 0.03) and DGel ± 95 % CIs decreased with increasing molecular weight (MW) from 7.1 to 5.1 (± 0.1-0.6) × 10-6 cm2 s-1. For the other 21 PFAS, linear regression models (DGel vs. MW; R2 ≥ 0.967) were used to estimate DGel ± 95 % CIs from 3.4 to 7.6 (± 0.2-1.0) × 10-6 cm2 s-1. Compared to their free water diffusivities, DGel values differed by a median of 6.5 % and first and third quartiles of 4.5 and 8.7 %, respectively. Error in DGel was propagated into CDGT for 5-36-day laboratory-scale DGT deployments, in which CDGT ± 95 % CIs for 20-31 of the 32 PFAS were indistinguishable from grab samples (sign test; α = 0.05). DGTs accurately captured time-weighted average PFAS aqueous phase concentrations at ∼1, 10, 100, and 200 ng L-1. This study demonstrates the utility of DGTs for PFAS quantitation at low ng L-1 levels and establishes methods for PFAS DGel determinations, error propagation into CDGT, and comparisons with grab sampling.
