PFAS Removal from Water: GAC, Ion Exchange and RO

What are PFAS and why are they so hard to remove?

Per- and polyfluoroalkyl substances (PFAS) are a family of thousands of synthetic chemicals built around a perfluorinated carbon chain. Each carbon carries fluorine atoms, and the carbon-fluorine (C-F) bond is the strongest single bond in organic chemistry, with a bond dissociation energy of roughly 485 kJ/mol. That thermodynamic stability is exactly why PFAS are called “forever chemicals”: nothing in the environment — microbes, sunlight, or conventional oxidants — supplies enough energy to cleave it, so the molecules persist for decades.

Structurally, most regulated PFAS are amphiphilic: a hydrophobic, oleophobic fluorinated tail paired with a hydrophilic head group (carboxylate in PFOA, sulfonate in PFOS). This surfactant character means PFAS accumulate at interfaces and resist the phase behaviour that conventional treatment relies on. Chain length is the master variable: long-chain species such as PFOA (C₈) and PFOS (C₈ sulfonate) are more hydrophobic and easier to capture, whereas short-chain replacements such as PFBA, PFBS and GenX (HFPO-DA) are more mobile and far harder to hold on a sorbent.

Why do biological and oxidation treatment fail?

Standard wastewater and drinking-water processes are designed to either mineralise organics (biology, advanced oxidation) or transform them. PFAS defeat both. Aerobic and anaerobic bacteria cannot access energy from the C-F bond, so PFAS pass through activated sludge essentially unchanged — and precursor compounds can even biotransform into regulated PFOA/PFOS, raising measured concentrations. Hydroxyl-radical advanced oxidation (ozone, UV/H₂O₂) has an oxidation potential adequate for most micropollutants but too low to strip fluorine from a fully fluorinated chain; the radicals are scavenged by background organics long before they touch the perfluoro backbone.

The practical consequence is that mainstream PFAS control is a separation-and-concentration problem, not a destruction problem. The three proven routes all move PFAS from a large dilute stream onto a small concentrated phase. Destruction technologies exist but currently sit downstream of that concentration step.

How does GAC adsorption remove PFAS?

Granular activated carbon removes PFAS by physical adsorption: the hydrophobic tail partitions onto the internal micropore surface while the charged head group interacts weakly with the carbon. Equilibrium loading is described by the Freundlich isotherm, which fits PFAS on GAC well across the trace concentration range.

Two effects dominate real performance. First, competitive adsorption: natural organic matter (NOM) occupies the same pores, depressing K_F and accelerating breakthrough relative to clean-water isotherms. Second, chain-length selectivity: short-chain PFAS break through first — sometimes within a few thousand bed volumes — so a bed still removing PFOS may already be leaking PFBA. Because full-scale kinetics are hard to predict, designers use rapid small-scale column tests (RSSCTs) on the actual water to scale breakthrough before committing to a media volume and changeout interval.

Worked example: GAC bed life and short-chain breakthrough

Consider a groundwater at 100 m³/h to be polished for PFAS on a lead-lag GAC train. The key sizing parameter is the empty bed contact time (EBCT), which fixes the carbon volume.

• Media volume: at EBCT = 10 min = 0.167 h, V_GAC = Q × EBCT = 100 × 0.167 = 16.7 m³ of carbon (per lead vessel).
• PFOS bed life: RSSCT indicates breakthrough at ~50,000 BV. Volume treated = 50,000 × 16.7 = 833,000 m³. At 100 m³/h that is 8,333 h ≈ 347 days, roughly annual changeout.
• Short-chain (PFBA) bed life: at ~10,000 BV, the same bed treats only 167,000 m³ = 1,667 h ≈ 70 days — five times shorter.

The lesson is that if the compliance driver is a short-chain PFAS, GAC changeout frequency (and carbon cost) rises sharply, and ion exchange or membranes often become more economical. Sizing must be pinned to the governing compound, not to PFOS alone.

How do ion exchange and NF/RO compare with GAC?

Selective anion-exchange (IX) resins target the anionic head group of PFAS directly, exchanging it for chloride on a quaternary-amine functional site. Purpose-built PFAS resins combine electrostatic attraction with a polystyrenic backbone that also engages the hydrophobic tail, giving very high capacity and much longer bed life than GAC — often 100,000+ bed volumes for long-chain PFAS — in a smaller footprint. They still break through faster on short-chain species and are sensitive to competing anions (sulfate, nitrate).

High-pressure membranes take a different approach: nanofiltration (NF) and reverse osmosis (RO) reject PFAS by size exclusion and charge (Donnan) repulsion rather than adsorption, so they are not consumed and reject both long- and short-chain PFAS. Well-designed RO achieves >99% rejection even of GenX; NF rejects long-chain PFAS strongly but can pass some of the smallest molecules. The trade-off is energy, pretreatment and a concentrate stream. See our note on PFAS-rejecting membrane systems and the sizing logic in our reverse osmosis system design guide.

Can PFAS be destroyed rather than just moved?

Because GAC, IX and membranes only concentrate PFAS, interest has shifted to destruction of the concentrated residual. Several routes are maturing:

• Foam fractionation exploits PFAS surface activity: sparging air through contaminated water draws the amphiphilic molecules to rising bubbles, producing a small foamate that concentrates PFAS by orders of magnitude — a concentration step, not destruction, but an efficient front end.
• Electrochemical oxidation at boron-doped diamond anodes generates sufficiently powerful surface oxidants to defluorinate PFAS, releasing fluoride; it works best on the concentrated foamate or reject rather than the dilute raw water.
• Supercritical water oxidation (SCWO) takes water above its critical point (>374 °C, >22.1 MPa), where organics and oxidant become fully miscible and PFAS mineralise to CO₂, water and fluoride within seconds — the most complete but most capital-intensive destruction route.

A realistic modern PFAS train therefore pairs a separation stage (GAC/IX/RO) with a concentration stage (foam fractionation) feeding a destruction stage, so that only a tiny high-strength stream needs the expensive oxidation.

What are the regulatory drivers and residuals obligations?

The regulatory ratchet is what makes PFAS removal from water a live engineering question rather than a research curiosity. In April 2024 the US EPA finalised the first federal drinking-water Maximum Contaminant Levels (MCLs): 4 ng/L (parts per trillion) each for PFOA and PFOS, 10 ng/L each for PFHxS, PFNA and GenX (HFPO-DA), plus a Hazard Index for mixtures. In Europe, the recast EU Drinking Water Directive (2020/2184) sets 0.10 µg/L for the sum of 20 PFAS. In the UK, the Drinking Water Inspectorate applies tiered guidance (a 100 ng/L action level for individual PFAS), while the Environment Agency is expanding PFAS monitoring across discharges and the water environment.

Residuals management is the obligation designers most often underestimate. Spent GAC and IX resin are PFAS-laden wastes requiring high-temperature reactivation or secure disposal; membrane reject is a low-volume, high-concentration liquor that cannot simply be returned to sewer. Each removal route trades a water problem for a residual problem, and that residual sits squarely inside your discharge and environmental permit obligations. For the wider UK consent framework, see our guide to effluent discharge standards in the UK.

Selecting a PFAS removal train

1. Characterise the PFAS profile. Speciate the water — long- vs short-chain, total concentration and the governing compound — plus NOM and competing anions that drive competition.
2. Fix the target. Identify the binding limit (EPA MCL, EU DWD, or UK/EA consent) and the specific PFAS that governs it; design to that compound, not to PFOS alone.
3. Run bench/RSSCT trials. Use rapid small-scale column tests for GAC and IX, and bench membrane rejection tests, on the actual water to get real breakthrough and rejection data.
4. Screen the technology. Match route to profile: GAC/IX for long-chain and moderate concentration; NF/RO where short-chain or very low limits dominate.
5. Design the residuals path. Plan reactivation or disposal for spent media and a management route for membrane reject before finalising the process.
6. Add destruction if warranted. Where residual volumes or liabilities justify it, concentrate by foam fractionation and destroy by electrochemical oxidation or SCWO.