PFAS mitigation is a drinking-water system-design challenge: lessons from contrasting architectures

Per- and polyfluoroalkyl substances (PFAS) comprise a large and diverse class of more than 14,000 synthetic compounds characterized by exceptionally strong carbon-fluorine bonds, which confer extreme chemical stability and resistance to environmental degradation¹. Their extensive use has led to widespread contamination of surface water, groundwater, and even rainwater on a global scale, typically at concentrations ranging from sub-ng/L to several hundred ng/L, with localized hotspots reaching mg/L in highly contaminated environments, such as industrially impacted rivers (e.g., the Xiaoqing River in China)^1,2,3,4. Intercepting PFAS from water supplies poses severe challenges to drinking-water safety and regulatory compliance, exacerbated by their well-documented adverse effects on human health and ecosystems^5,6,7.

In response, regulatory initiatives led by the United States Environmental Protection Agency (U.S. EPA) and the European Chemicals Agency (ECHA) have driven a worldwide industrial shift away from long-chain PFAS⁸. These legacy compounds include perfluoroalkyl carboxylic acids (PFCAs) with ≥C₇ fully fluorinated carbons, such as perfluorooctanoic acid (PFOA, C₈) and perfluoroalkane sulfonic acids (PFSAs) with ≥C₆ carbons, such as perfluorooctanesulfonic acid (PFOS, C₈). These compounds are stringently regulated under the U.S. Safe Drinking Water Act and the European Union Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) Regulation^9,10. In their place, industries have increasingly adopted short-chain alternatives, such as PFCAs with ≤C₆ carbons, including perfluorobutanoic acid (PFBA, C₄) and perfluorohexanoic acid (PFHxA, C₆), and PFSAs with ≤C₅ carbons, such as perfluorobutanesulfonic acid (PFBS, C₄)⁹. These short-chain alternatives are highly mobile and extremely persistent, leading to widespread water contamination and poor removal by conventional treatment processes^11,12. Compounding this challenge, ultra-short-chain PFAS, such as trifluoroacetic acid (TFA, C₁) and perfluoropropanoic acid (PFPrA, C₃), are increasingly detected as both transformation products and direct emissions¹³. They have recently become a growing concern because they are often excluded from routine analytical measurements, despite dominating the PFAS profile in many samples. Moreover, conventional water-treatment processes cannot effectively remove TFA from water.

Globally, PFAS management remains dominated by conventional treatment in groundwater- and surface-water-based systems¹⁴, while only a limited number of countries with high desalination dependence rely primarily on reverse osmosis (RO) as their central PFAS barrier¹⁵. However, conventional treatment processes generally fail to provide effective PFAS removal¹⁶. While desalination using RO effectively removes PFAS from treated drinking water¹⁷, it transfers PFAS contamination burden to the environment through brine discharge rather than eliminating the total mass¹⁸. The same system-level framework applies to regions relying on contaminated groundwater, PFAS-affected surface waters, wastewater reuse, or water sources influenced by industrial emissions, which reinforces that PFAS mitigation strategies must be selected according to the complete contaminant pathway, including source, interception point, residual generation, and access to destruction or disposal infrastructure.

This Comment article, developed under a German-Israeli collaboration and supported by the science ministries of both countries (BMBF-MOST), argues that PFAS mitigation should be understood as evidence-based examples for a system-design challenge rather than solely a treatment problem. Using Germany and Israel as contrasting illustrative cases, we show how water-source selection and treatment architecture determine where PFAS are intercepted, where residual burdens accumulate, and which mitigation strategies are technically and environmentally realistic. The broader lesson is that the same regulatory target can lead to very different management pathways depending on the design of the drinking-water system.

PFAS treatment approaches cannot be evaluated in isolation from the systems in which they are deployed. Their practical suitability depends not only on removal performance but also on the dominant water source, existing infrastructure, target PFAS composition, regulatory framework, and the characteristics of the residual stream generated. Powdered activated carbon (PAC), granular activated carbon (GAC), ion-exchange resins (IERs), and dense membrane processes such as nanofiltration (NF) and RO provide different combinations of operational maturity, PFAS selectivity, infrastructure footprint, and residual-management demands. Accordingly, the central question is not which technology achieves the highest removal under isolated conditions, but where PFAS can be intercepted most effectively within a specific drinking-water network and how the resulting residual burden is managed downstream.

Germany and Israel illustrate two contrasting system configurations. Germany represents regions with relatively abundant freshwater resources, where drinking water is derived primarily from groundwater (74.1%) and surface water (20.7%), with an additional 5.2% supplied indirectly through bank filtration or artificially groundwater recharged¹⁹. While natural subsurface filtration contributes substantially to freshwater purification^20,21, conventional treatment trains were not designed for effective PFAS removal²². Tightened limits under the amended German Drinking Water Ordinance, aligned with the broader EU Drinking Water Directive framework, now oblige water suppliers at an increasing number of PFAS-affected sites to integrate additional treatment processes or abandon impacted raw water sources. As a result, effective PFAS management relies primarily on advanced add-on treatment processes that can be integrated into existing infrastructure.

Conversely, Israel represents regions that suffer from water scarcity, in which seawater RO has become a central component of the national drinking-water supply^23,24,25,26. This configuration fundamentally alters Israel’s PFAS management, as reported removal efficiencies for a broad range of PFAS (C₄-C₁₀) are typically on the order of 98–99%²⁷, effectively limiting consumer exposure in finished water. Rather than relying primarily on retrofitted capture within conventional plants, a high-pressure membrane provides an effective barrier to remove salts, trace organics, and most PFAS from source water before distribution. The contrast between the two countries therefore reflects different management philosophies: countries with abundant water sources that rely predominantly on adsorption-based treatments must introduce selective PFAS capture into existing groundwater- and surface-water-treatment infrastructure, whereas countries with water scarcities that rely on RO systems already operate a centralized membrane barrier but must address the liquid residuals produced by that barrier. Detailed regulatory milestones and monitoring frameworks for both countries are provided in Supplementary Discussion 1 and Supplementary Discussion 2.

Activated carbon (AC) is among the most technologically mature and operationally familiar approaches for integrating PFAS removal into conventional drinking-water treatment. AC is already widely used in waterworks²⁸ and can be incorporated in powdered or granular form at comparatively low cost^29,30. PAC is usually added to water at various locations in the water treatment train, and it is then removed by sedimentation or filtration³¹, whereas GAC can be operated in a downflow mode as a fixed-bed barrier³⁰. Their compatibility with established infrastructure makes AC particularly relevant to systems in which the construction of a new membrane-based treatment train would be technically or economically disproportionate. Full-scale experience further illustrates the marked selectivity of GAC for different PFAS classes. At the Rauental waterworks, conventional GAC filters treating PFAS-contaminated raw water exceeded the German PFAS-20 limit of 100 ng/L after fewer than about 20,000 bed volumes, and even with PFAS-4 as the treatment target, replacement was still required after roughly 20,000–30,000 bed volumes³². This limited run time was driven primarily by the early breakthrough of short-chain PFCAs (e.g., PFBA and PFPeA), while longer-chain PFASs such as PFOA remained more strongly retained. GAC should therefore be regarded as a robust and practical compliance tool within existing systems, but not as a comprehensive solution where short-chain and highly mobile PFAS dominate.

Ion exchange resins (IERs), particularly strong-base anion exchange resins (AERs), are increasingly adopted to complement or replace GAC in PFAS treatment³³. In systematic batch and column experiments with different polystyrene (PS)‑based AERs with quaternary ammonium functional groups and polyacrylic/polyacrylamide (PA)‑based AERs, short‑chain PFCA (down to C₁-C₃) and selected sulfonates were removed to a relevant extent, but performance proved highly sensitive to competing inorganic anions, natural organic matter, empty bed contact time and filter velocity³². Furthermore, regeneration yields are strongly resin- and compound-specific. While nitrate-based regenerants achieve high recovery for a range of PFCA and PFSA³⁴, NaOH-only regeneration, commonly used for non-PFAS-specific AERs, often remains incomplete³². This results in spent resins typically being disposed of by incineration, raising environmental and cost concerns.

The comparison between GAC and IER illustrates that improved short-chain PFAS capture does not automatically produce a more feasible system-level solution. IERs may improve affinity for selected mobile PFAS and reduce the required treatment footprint, but their performance is more sensitive to water composition, and regeneration can create concentrated liquid wastes requiring further treatment. GAC is more established and operationally familiar, but short-chain breakthrough may require frequent media replacement. This was well established by Riegel et al.³⁵, who quantified breakthrough for PFBA, PFPeA, and PFHxA across six AER and benchmarked them against GAC³⁵. He has presented data that show that recently developed PFAS-selective IER (especially styrene-divinylbenzene copolymers operated with tailored ethanol-salt regenerants) can match or outperform GAC for short‑chain PFAS under conditions where multiple regeneration cycles are feasible, and PFAS are dominated by C₄-C₇ carboxylates. GAC still tends to be superior or more economical for long-chain PFAS and for single-use operation without regeneration. Sequential application may therefore be advantageous in selected systems, with GAC reducing competing background constituents before ion exchange captures the remaining PFAS. The suitability of this arrangement nevertheless depends on local water chemistry, regulatory approval, and the availability of a residual-treatment pathway.

Dense membrane processes provide a different treatment philosophy from adsorption-based capture. Nanofiltration (NF) and RO can reject a broad range of PFAS while simultaneously removing salts and other dissolved contaminants. Their role is therefore particularly significant where membrane treatment is already embedded within the water-supply system or where several water-quality objectives justify the energy and infrastructure requirements of high-pressure filtration.

In regions that rely primarily on adsorption-based treatment, such as Germany, membrane treatment plays a secondary role in drinking-water production, although its application is increasing. Ultrafiltration (UF) is applied mainly for surface waters, and membrane processes account for less than 10% of total drinking water treatment in Germany³⁶. Dense NF and RO processes are also increasingly applied, with approximately 90 plants and an installed treatment capacity exceeding 10,000 m³/h^37,38. The implementing RO specifically for PFAS removal generally represents an additional treatment investment rather than the use of an existing national-scale barrier. Nevertheless, the number of plants is expected to grow further as expectations regarding drinking water quality rise and new, previously unknown or neglected substances are detected in raw water³⁸. The Sandweier waterworks illustrates both the effectiveness and the system-level limitation of this approach. An RO system was implemented at the Sandweier waterworks to treat contaminated groundwater³⁹. PFAS are not detectable in the permeate (<1 ng/L), and the overall PFAS rejection exceeds 2.5 log units⁴⁰. However, the process generated a concentrate containing PFAS at concentrations ~5-fold higher than those in the feed³⁹. Direct discharge of this concentrate was not authorized, requiring the installation of an additional GAC system for concentrate treatment⁴⁰. This example demonstrates that high permeate quality does not by itself resolve the PFAS burden, membrane implementation also requires a technically and legally acceptable concentrate-management pathway.

In water-scarce regions, such as Israel, by contrast, drinking-water treatment is dominated by seawater RO, which has become the backbone of the national water supply^23,24,25,26. This configuration allows effective removal of many PFAS before water enters the distribution network, but it does not destroy them. RO instead concentrates PFAS in the reject stream⁴¹. Seawater desalination brine typically accounts for approximately 50% of the feedwater volume⁴², and demands roughly 3-4 kWh/m³ for the treatment⁴³. Consequently, the system-level challenge is not limited to maintaining finished-water quality but extends to the cumulative PFAS burden discharged in large-volume brine streams⁴⁴. Groundwater remains a strategic reserve in Israel to support resilience during drought or infrastructure failure. It also requires far less energy than seawater desalination, with extraction alone consuming about 0.8–1.2 kWh/m³ ^45,46, and conventional treatment an additional 0.1–0.4 kWh/m³ ^47,48. This results in PFAS management being consequently less about upgrading finished-water treatment across the national system and more about managing two specific challenges: contaminated groundwater sources outside the desalination network and PFAS-enriched brines generated by large-scale desalination. However, groundwater sources affected by PFAS fall outside the principal desalination barrier and may require targeted remediation, adsorption, or other add-on processes⁴⁹. Thus, even within a desalination-oriented national system, different source waters may require different interception strategies.

The comparison demonstrates that NF and RO should not be considered universally superior to adsorption solely because they provide broader PFAS rejection. In conventional groundwater systems, their installation may require substantial energy, infrastructure modification, and concentrate treatment. In desalination-based systems, the membrane barrier already exists, making PFAS rejection an additional benefit of the established treatment architecture; however, the resulting environmental burden is transferred to high-volume brine. The feasibility of membrane treatment therefore depends on both its position within the existing system and the availability of an appropriate concentrate-management strategy.

The contrasting German and Israeli configurations demonstrate that the same PFAS-compliance objective can require different mitigation pathways depending on drinking-water system architecture (summarized in Fig. 1). Groundwater-based systems such as Germany’s favor modular add-on capture, whereas desalination-based systems such as Israel’s already incorporate RO as a PFAS barrier, shifting the principal challenge toward management of PFAS-enriched brine. Accordingly, PFAS treatment technologies selection should consider the complete PFAS pathway, including where contamination can be intercepted, compatibility with existing infrastructure, the form and volume of residuals generated, and the availability of suitable regeneration, destruction, or disposal pathways. Germany and Israel therefore illustrate a broader principle: drinking-water architecture determines not only which PFAS mitigation strategy is operationally feasible, but also where the residual contaminant burden ultimately accumulates.

The treatment strategies discussed above protect drinking-water consumers primarily by separating PFAS from the treated water rather than destroying them. Consequently, treatment shifts the PFAS burden into secondary residuals, including PFAS-loaded spent adsorbents, regenerant solutions, and membrane concentrates. The unresolved challenge is therefore beyond achieving regulatory PFAS limits, but also closing the PFAS mass balance across the treatment system by shifting from phase-transfer separation to a destruction pathway.

For adsorption-based treatment, the primary downstream challenge is the management of PFAS-loaded solid media. GAC remains one of the most established PFAS capture technologies, while sequential GAC-IER treatment may improve removal across a broader range of PFAS. As Rückbeil et al.⁵⁰ demonstrated, in such configurations, GAC can reduce competing background constituents before IER captures remaining PFAS, including more mobile species⁵⁰. However, the limitations of this approach are that IER is more expensive than AC, as Marshall’s thesis⁵¹ estimated planning-level costs for drinking water treatment using single-use (non-regenerated) resin at €0.07–€0.18 per m³, with higher costs incurred under stricter breakthrough criteria⁵¹. Regulation can also determine the fate of downstream solid residuals. Once PFAS have been captured, there is also the constraint of waste management. PFOS, PFOA and perfluorohexanesulfonic acid (PFHxS) are classified as persistent organic pollutants (POPs) and are listed in Annex IV of the European POP-regulation with very low concentration limits for wastes: 50 mg/kg for PFOS and its derivatives and 1 mg/kg for PFOA and PFHxS (plus 40 mg/kg for their related substances)⁵². Once these thresholds are exceeded, spent GAC is classified as POP‑contaminated waste and must be treated by one of the disposal operations listed in Annex V (e.g. D10 incineration on land), which restricts options for standart regeneration and drives up costs⁵³.

Until recently, the lack of robust, standardized methods to quantify PFAS on spent AC made it difficult to demonstrate compliance with POP thresholds and to distinguish between GAC that could be safely reactivated and GAC that required direct high-temperature disposal⁵⁴. Improved analytical approaches based on liquid chromatography-tandem mass spectrometry (LC-MS/MS), quantitative fluorine-19 nuclear magnetic resonance (¹⁹F NMR), and fluorine mass balances allow more reliable determination of PFAS loadings on exhausted carbon and tracking of defluorination during thermal treatment⁵⁵. Full-scale and pilot studies of thermal reactivation of PFAS-laden GAC have reported destruction and removal efficiencies exceeding 99.9% for regulated PFAS, with fluorine mass balances typically in the range of 60–90%, indicating extensive conversion of PFAS to inorganic fluoride species⁵⁶. PFAS-laden GAC may be thermally reactivated only if complete destruction of the POP content can be demonstrated in the overall process and the reactivated carbon complies both with POP trace-level requirements and drinking-water purity criteria; otherwise, the spent carbon must be treated as POP waste and sent to dedicated incineration routes⁵³.

Capturing PFAS on a solid medium can facilitate thermal treatment because the contaminant burden is transferred from a large volume of drinking water into a comparatively small mass of material. Consequently, the appropriate evaluation criterion is not adsorption capacity alone, but the complete pathway from PFAS capture to residual treatment and verified destruction. The critical next step is therefore not only to increase removal efficiency but also to scale up residual‑management pathways that combine demonstrable PFAS destruction with the minimization of secondary waste streams and resource use under the POP framework.

Membrane-based separation produces a fundamentally different residual-management challenge. NF and RO can achieve high PFAS rejection, but the retained compounds remain in the membrane concentrate. Reported PFAS concentrations in Mediterranean Sea water are typically in the hundreds of pg/L, with values of 246–515 pg/L⁵⁷. Israeli coastal desalination capacity is approximately 587 million m³ of drinking water/year⁴². Accordingly, although concentrations entering seawater RO may be low, the large treated volumes result in PFAS being transferred continuously to desalination brines that are discharged to the Mediterranean Sea^42,58. The system-level priority is therefore not simply to improve front-end PFAS rejection, but to develop downstream treatment pathways that can be coupled to large-scale desalination infrastructure. This need has recently been highlighted in US-Israel collaborations⁵⁹, yet such solutions remain in their early stage of development.

For such aqueous residuals, pre-concentration becomes an important design step because destruction technologies are generally more feasible when the volume requiring treatment is reduced. Membrane-based brine concentration has already been demonstrated at pilot scale. Pre-concentration using NF/RO has been demonstrated by Abdalrhman et al.⁶⁰ at pilot scale, where an NF-RO-high-pressure NF system concentrated seawater RO brine from ~75–78 g/L to 220–251 g/L NaCl with stable operation over 80 days and a specific electricity consumption of 111–123 kWh per ton of NaCl⁶⁰. Rather than treating membrane concentration as an independent technology, its system-level role is therefore to condition a large-volume residual stream for subsequent destruction.

Among the destruction options considered for concentrated aqueous PFAS wastes, supercritical water oxidation (SCWO) is comparatively advanced^61,62. SCWO operates above 374 °C and 22.1 MPa, where organics become soluble, oxygen is fully miscible, and salts are insoluble⁶¹. In this supercritical phase, fluorinated organics are oxidized to CO₂ and fluoride. Krause et al.⁶¹ treated the brines of three commercial SCWO units at ~590 °C and ~24 MPa with residence times of 10–68 s, achieving >99% reduction in total targeted PFAS in dilute aqueous film-forming foam and showing fluoride formation consistent with extensive mineralization⁶¹. Scheitlin et al.⁶² optimized the Battelle PFAS Annihilator at 600 °C, at ~24 MPa as well, and residence times on the order of tens of seconds, demonstrating >99.9998% reduction and 70-73% recovery of fluorine as inorganic fluoride, with steady-state effluent PFAS < 70 ng/L⁶². Despite this high destruction efficiency, SCWO remains capital- and energy-intensive, with an estimated electricity demand of approximately 1.1 × 10⁵ kWh/kg of PFAS destroyed⁶³, reinforcing the importance of reducing residual volume before treatment.

Electrochemical oxidation (EO) represents another potential pathway for concentrated liquid streams, as a more moderate-energy alternative. Laboratory and pilot studies have shown 80–98% destruction of long-chain PFAS in concentrated RO brine for more efficient mineralization⁶⁴, with relatively modest reported energy consumption (0.07-0.16 kWh/m³ in treatment-train configurations). The high ionic strength of brine may enhance mass transfer, potentially reducing the need for additional electrolytes and further reducing cost. However, byproduct formation under these conditions requires further characterization⁶⁴.

Plasma-based treatment is also being investigated for concentrated PFAS-containing liquids. In the reported configuration, PFAS are first concentrated using IER and then degraded by plasma. After 6 h of operation, up to 97% degradation of total PFAS has been demonstrated⁶⁵. Nevertheless, plasma-based systems are still at an early stage of practical deployment. Some short- and ultrashort-chain substances (e.g., perfluoropentanoic acid and PFBA) may transiently increase in concentration before being further degraded with extended exposure. Plasma-based systems are also known to be energy-intensive, consuming ~12 × 10³ kWh/kg of PFAS destroyed, with estimated operating costs of €0.24 per liter⁶⁵.

Taken together, these approaches present a second broader design principle: destruction technologies for aqueous PFAS residuals should not be evaluated independently of the separation and concentration stages that precede them. A process capable of high destruction efficiency may remain impractical if it must treat the entire high-volume residual stream generated by the drinking-water system. Conversely, upstream concentration can reduce the volume entering the destructive process but creates additional energy, scaling, and residual-management requirements. System design must therefore optimize the complete treatment chain rather than the efficiency of an individual unit operation.

When comparing groundwater-oriented conventional infrastructure and desalination-oriented infrastructure, neither pathway is inherently universal or superior; rather, each reflects the residual stream created by the existing drinking-water architecture. The proposed strategies for handling PFAS residuals within each infrastructure (outlined in Fig. 2) emphasize that the central challenge is PFAS transferring from drinking water into residual streams. The broader lesson is that PFAS mitigation should progress from removal-oriented treatment design toward mass-balance-oriented system design. This transition from phase transfer to system closure represents the central engineering challenge for the next generation of PFAS management strategies. A ministerial cooperation between Germany and Israel under the BMBF-MOST framework, aimed at identifying system-specific solutions while sharing knowledge from the established drinking-water treatment infrastructures on which each country relies, is described in Supplementary Discussion 3.

Water-source characteristics and existing treatment architecture first determine where PFAS can realistically be intercepted: conventional groundwater- and surface-water systems are generally more compatible with add-on adsorption processes, whereas desalination-based systems already incorporate RO as a centralized separation barrier. PFAS composition further constrains this choice, as GAC is more effective for longer-chain PFAS but is limited by earlier breakthrough of short- and highly mobile compounds, while IER may improve the capture of selected mobile PFAS but remains sensitive to water composition and can generate concentrated regenerant wastes. NF and RO provide broader PFAS separation but require greater energy input and transfer PFAS into concentrated liquid residuals. However, the global challenge is not simply to maximize PFAS removal from finished water, but to connect separation with a feasible pathway for residual management and destruction. Adsorption-based treatment transfers PFAS to spent solid media that may be directed toward thermal treatment or reactivation, whereas membrane-based treatment transfers the burden to aqueous concentrates that may require further concentration before destruction. Emerging destruction approaches, including SCWO, electrochemical oxidation, and plasma-based treatment, demonstrate the potential for treating concentrated liquid residuals, although their applicability is constrained by energy demand, technological maturity, and residual-stream characteristics. Consequently, systems with different source waters, treatment infrastructures, regulatory requirements, and available resources may require different pathways to achieve the same drinking-water objective. The globally transferable lesson is therefore not the selection of a particular technology, but the need to evaluate the complete PFAS pathway-from interception and concentration to residual management, destruction, and verification of system closure-rather than judging treatment success solely by finished-water removal efficiency.