Beyond the isotherm: Limits in prediction of PFAS breakthrough behaviour in continuous flow column

Batch kinetic isotherm and continuous flow column experiments were performed for five per- and polyfluoroalkyl substances (PFAS). Coefficient of determination (R 2 ) was higher for Pseudo-Second-Order (PSO, 0.92) than Pseudo-First-Order (PFO, 0.88) kinetic model, suggesting that PFAS removal mechanism was probably based on chemisorption. Freundlich constant (K F ) representing the adsorption capacities revealed that anion exchange resin (AER) consistently outperformed granular activated carbon (GAC) for all tested PFAS. The adsorption capacity of GAC ((mg/g)/(mg/L) 1/n ) increased with increasing chain length from 0.19 for Perfluoropentanoic acid (PFPeA) to 4.94 for perfluoroundecanoic acid (PFUnDA). AER was seen to remove both short and long-chained PFAS effectively. Consistent with that, a clear correlation was observed between the difference in adsorption capacities (ΔK F ) of the adsorbents and PFAS chain length in that AER:GAC adsorption capacity ratio maintained the following order: PFPeA (46) > Perfluorooctanoic acid, PFOA (23.5) > Perfluorononanoic acid, PFNA (8.4) > Perfluorodecanonic acid, PFDA (5.2) > PFUnDA (2.8). This confirms higher adsorption of longer chain PFAS on GAC. In the continuous flow column tests, for GAC, shortest chain PFPeA exhibited the earliest breakthrough at ~7000 bed volumes (BV, day 36), followed by PFOA at ~10,000 BV (day 50). In contrast, PFNA, PFDA, and PFUnDA showed delayed breakthrough at around ~12,000 BV (day 59). For AER, PFPeA again exhibited the earliest breakthrough at ~10,000 BV (day 52), followed by PFOA at ~11,000 BV (day 56) PFDA ~12,000 BV (day 61), PFNA ~13,000 BV (day 66) and PFUnDA~14,000 BV (day 72). The results show that n and K F values measured through batch isotherm tests significantly underestimate the actual breakthrough period observed in continuous flow column experiments. • Adsorption of PFAS on GAC is a function of chain length. • AER consistently outperformed GAC for all PFAS. • Efficiency gap between GAC and AER decreased with increasing PFAS chain length. • Isotherm predicted chain-length dependent trends as in continuous flow columns. • Isotherm underpredicted the service life of the fixed bed continuous flow column.