Front. Environ. Eng. Frontiers in Environmental Engineering Front. Environ. Eng. 2813-5067 Frontiers Media S.A. 1766573 10.3389/fenve.2026.1766573 Original Research A structured forensic framework for PFAS source differentiation under target-only analytical constraints Zenobio et al. 10.3389/fenve.2026.1766573 Zenobio Jenny E. 1 Conceptualization Formal Analysis Methodology Writing – original draft Writing – review and editing Pazoki Faezeh 2 Writing – original draft Writing – review and editing Forsberg Adam 3 Formal Analysis Methodology Writing – review and editing Chiang Sheau-Yun Dora 3 * Conceptualization Methodology Writing – review and editing Jacobs, Irvine, CA, United States Jacobs, Toronto, ON, Canada Jacobs, Atlanta, GA, United States Correspondence: Sheau-Yun Dora Chiang, dora.chiang@jacobs.com 23 02 2026 2026 5 1766573 12 12 2025 28 01 2026 02 02 2026 Copyright © 2026 Zenobio, Pazoki, Forsberg and Chiang. 2026 Zenobio, Pazoki, Forsberg and Chiang https://creativecommons.org/licenses/by/4.0/ This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

Differentiating overlapping sources of per- and polyfluoroalkyl substances (PFAS) remains a central challenge in environmental forensics, particularly where investigations rely on targeted analytical datasets. Here, we present a tiered PFAS fingerprinting framework designed to extract source, process, and transport information using only target analytes. The framework integrates multiple, complementary lines of evidence, including compound-level concentrations, class- and carbon-number-resolved composition, diagnostic ratios, isomer distributions, precursor-product relationships, multivariate clustering, and geospatial pattern analysis, to support defensible source differentiation under data-limited conditions. The framework is demonstrated using groundwater datasets collected at two time points (2018 and 2024) from a complex industrial setting with overlapping PFAS inputs. Application of the framework resolves distinct PFAS mixture archetypes that reflect differences in manufacturing era, formulation chemistry, and hydrologic context. Identified profiles include sulfonate-rich mixtures consistent with electrochemical fluorination-era inputs, telomer-associated industrial mixtures characterized by fluorotelomer sulfonates and carboxylates, and short-chain-enriched profiles influenced by wastewater-related transport and mixing. Temporal evaluation shows changes in precursor abundance and terminal perfluoroalkyl carboxylic acids, between sampling events, while diagnostic ratios and isomer patterns provide additional temporal context where quantifiable. Unsupervised clustering independently corroborates compositional similarity and hydraulic connectivity among site domains. Together, these results indicate that target-only PFAS datasets can support forensic interpretation when multiple, complementary analytical metrics are evaluated in a structured framework. The approach outlines an analytical structure that could assist PFAS investigations where source histories are complex and compound coverage is limited.

 clustering analysis fingerprint PFAS sources PFOA PFOS source differentiation The author(s) declared that financial support was not received for this work and/or its publication. section-at-acceptance Water, Waste and Wastewater Engineering Introduction

Differentiating sources of per- and polyfluoroalkyl substances (PFAS) is increasingly critical for environmental forensics, liability allocation, remedial planning, and interpretation of complex site datasets. Many contaminated locations contain multiple overlapping release mechanisms, including historical aqueous film forming foam (AFFF) use, industrial discharges, landfill leachate, wastewater treatment system, and atmospheric deposition. Each produces distinct compositional patterns (Charbonnet et al., 2021; Joseph et al., 2025; Ruyle et al., 2021b; Benotti et al., 2020; Langberg et al., 2022). Resolving these contributions is essential for constructing accurate conceptual site models (CSMs) and supporting regulatory decisions.

Most PFAS monitoring programs in the U.S. depend solely on targeted PFAS analysis, dominated by U.S. Environmental Protection Agency (EPA) Method 1633, which quantifies forty (40) PFAS across a narrow set of perfluoroalkyl carboxylic acids (PFCAs), perfluoroalkyl sulfonic acids (PFSAs), selected fluorotelomerization (FT)-derived acids, and a small number of sulfonamido precursors (U.S. EPA, 2024). Within this constrained analyte list, non-target analysis (NTA) and suspect screening derived largely from high-resolution mass spectrometry (HRMS), provide access to hundreds of additional PFAS not captured in routine monitoring (Joerss and Menger, 2023; Liu et al., 2019). These NTA have not been standardized but widely used in research allowing for identification of proprietary or formulation-specific compounds, and enabling semiquantitative characterization of precursor families and intermediates that are relevant for PFAS fingerprinting (Joseph et al., 2023; Liu et al., 2019; Strynar et al., 2023). In practice, however, environmental investigations rarely have access to this level of analytical resolution. Within this constrained analyte list, we believe some source-relevant patterns remain observable, including PFSA-rich profiles typical of legacy AFFF formulations (Malik et al., 2024; Annunziato et al., 2020), PFCA-dominated patterns associated with specific industrial applications, and detectable FT intermediates that may indicate telomer-based products or precursor transformation (Strynar et al., 2023; He et al., 2022). Short-chain PFCAs can also serve as measurable indicators of precursor oxidation in groundwater or wastewater systems (Strynar et al., 2023; Eriksson et al., 2017). When considered alongside site-specific hydrologic information, infrastructure layout, release history, and geochemical conditions, these target-only patterns may provide context for interpreting potential PFAS source classes (Charbonnet et al., 2021; Joseph et al., 2023).

The objective of this paper is to introduce a PFAS source-differentiation framework tailored to the analytical realities of contemporary environmental investigations. This framework evaluates the extent to which meaningful differentiation can be achieved using only the target analytes while also clarifying the types of questions that cannot be resolved without broader HRMS, suspect screening, or NTA capabilities. This work establishes a scientifically defensible and practically implementable foundation for PFAS source differentiation under current constraints. The framework is illustrated to show how target-based PFAS compositional information can be interpreted within a structured analytical context.

 Methods

The PFAS forensic framework is developed based on the assumptions that only target PFAS datasets are available for evaluating source differentiation. Practical constraints relevant to target-PFAS fingerprinting, including limitations are discussed in Supplementary Section S1. The framework approach is organized into six tiers that integrate concentration magnitude, compositional structure, chain-length distribution, diagnostic ratios, spatial patterns, temporal evolution, and multivariate similarity (Table 1). Details of all computational steps, including data preprocessing, normalization, clustering parameters, ordination settings, and geostatistical methods, are provided in Supplementary Section S2.

Structured multi-tier framework for PFAS forensic fingerprinting under target-only datasets.

Tier Evaluation focus Key questions Primary metrics/Lines of evidence Required inputs Typical outcomes/Decisions
Tier 1 – CSM data gathering Construct PFAS-focused CSM What sources are plausible?Where and when could PFAS enter the system?How does PFAS move through hydrostratigraphy and infrastructure? Site history; PFAS-use chronology; known AFFF/industrial/wastewater activities; hydrostratigraphy; flow regime; release pathways Facility records; maps; cross sections; prior investigations Candidate source classes and entry points defined; working hypotheses established
Tier 2 – Compositional and concentration screening
Bulk impact and background Total PFAS Is the location background, downgradient, or source-proximal?Where is PFAS mass concentrated? ΣPFAS in media; comparison with regional + local background; concentration gradients Concentration dataset; background values; CSM flow context Locations categorized as background, downgradient receptors, or potential source/near-source zones
PFAS class Identify class-level composition Which classes dominate?What do patterns imply about manufacturing era and use domain? Relative contributions of PFSAs, PFCAs; telomer-era acids; ECF-derived sulfonamides; replacement ether/short-chain PFAS. Speciation data; knowledge of manufacturing timelines Broad source categories constrained (legacy ECF/AFFF, industrial PFCA, wastewater/landfill, modern replacements)
Chain-length patterns Carbon-number distributions Is the profile long-chain–rich or short-chain–dominated?Is enrichment source- or transport-driven? Long- vs. short-chain PFSA/PFCA fractions; enrichment trends along flow paths Speciation by carbon number; flow directions Legacy/older vs. modern/transform-derived patterns distinguished; source-proximal vs. downgradient zones resolved
Precursor dynamics Infer precursor contribution Do PFCAs reflect FT-precursor degradation?Do patterns constrain production era? PFPeA–PFHxA–PFHpA enrichment; presence/absence of FTS; expected terminal-product distributions from precursor oxidation PFCA distributions; manufacturing timelines; CSM temporal context FT precursor influence identified; degradation extent inferred; ECF vs. FT release windows constrained
Linear/Branched isomers Manufacturing-process resolution Do isomer patterns indicate ECF or FT production?Are isomers evolving with transport distance? Branched/linear fractions for PFOS, PFHxS, PFOA; attenuation differences (branched > linear mobility) Isomer-resolved data; production-era knowledge; flow pathways ECF-era vs. FT-derived inputs separated; isomer gradients used to refine plume maturity and transport distance
Tier 3 – Diagnostic ratios Composition-normalized indicators Do abundance ratios support specific source classes or mixture structures? PFOS/PFHxS (AFFF formulation variability); PFOS/PFOA (sulfonate vs. carboxylate dominance); ΣPFAS ≤ C6/ΣPFAS (short-chain enrichment from source, precursor turnover, or transport) Ratio calculations from measured analytes Differentiation of AFFF, industrial, wastewater/landfill, and transformed mixtures; recognition of mixing and non-source processes shaping profiles
Tier 4 – Spatial pattern analysis Horizontal + vertical structure Do PFAS patterns evolve systematically with distance or depth?Are plumes discrete or mixing? Upgradient–downgradient transitions; chain-length shifts; vertical alignment vs. stratification; lateral convergence/divergence Spatially referenced dataset; hydraulic gradients Transport pathways delineated; source zones distinguished from migrated signatures; mixed-plume regions identified
Tier 5 – Temporal trends Multi-round trends Are concentrations or compositions changing?Are new PFAS emerging? Directional concentration changes; PFCA increases indicating precursor turnover; appearance of replacement PFAS; seasonal patterns in short-chain PFAS. Multi-year/multi-season datasets Active vs. historical inputs distinguished; precursor activity confirmed; timing of new/recent releases constrained
Tier 6 – Multivariate clustering Resolve mixture structure statistically Do samples group into chemically coherent clusters?Are clusters spatially or temporally organized? Hierarchical clustering, k-means, PAM; multidimensional ordination (multidimensional scaling/partitioning around medoids); cluster stability; indicator analytes for each group Matrix of speciated PFAS across samples; spatial + temporal attributes Identification of discrete mixture types; separation of single-source vs. mixed-source groups; resolution of complex plumes not distinguishable via univariate metrics
Example datasets to demonstrate target PFAS forensic framework

The framework is demonstrated using an industrial scenario representative of conditions commonly encountered at facilities with potential PFAS inputs. The setting includes three operational domains with distinct activity histories relevant to PFAS occurrence (Supplementary Figure S1). The first domain represents a storage-facility area affected by a large fire in the late 1980s. The area experienced a period of inactivity following the event. The second domain represents an adjacent industrial corridor that has operated since the 1970s and includes industrial activities, where fluorinated surface-treatment chemistries may have been used. The third domain represents an onsite wastewater treatment plant (WWTP) that has been operational since the 1980s and receives industrial wastewater influent. Groundwater samples were collected in 2018 and 2024 and analyzed for target PFAS enabling comparative evaluation of spatial patterns and temporal changes in PFAS composition.

 Tier 1 – Data gathering for assembling conceptual site model

Hydrostratigraphy and groundwater flow dynamic data that control the migration and attenuation of PFAS in the environment were gathered. Based on this information, potential PFAS sources were assessed, including legacy ECF-related AFFF release history associated with the accidental event, industry-era releases, and wastewater-associated contributions. Other nearby PFAS source categories outside of property boundaries, such as routine fire-training activities, aircraft hangars, metal-plating operations, sewer mains or pump stations, landfills, or storage units, should be considered into PFAS source inventory. Where present, such features would typically be incorporated into a PFAS-specific CSM. Supplementary Figure S2 in the SI summarizes the introduction, production periods, and phase-out of major PFAS classes relevant to this study, including ECF-derived sulfonates and fluorotelomer-based chemistries. This timeline provides historical context for interpreting compound-specific patterns observed in groundwater.

 Tier 2 – Compositional and concentration screening of target analytes Total PFAS concentrations

Total PFAS concentrations (ΣPFAS) were evaluated to provide a first-order screening of spatial variability across the site and to contextualize observed concentrations relative to regional groundwater background. Measured ΣPFAS values were compared to published Minnesota ambient groundwater background concentrations (Supplementary Table S1) reported by the Minnesota Pollution Control Agency (MPCA, 2024) as an example on how to distinguish background-consistent conditions from locations exhibiting elevated PFAS levels. This comparison was used to support identification of areas potentially influenced by site-related PFAS inputs versus locations representative of regional background. Spatial distributions of ΣPFAS were visualized using scaled pie charts, with pie area normalized to total PFAS concentration to facilitate comparison across sampling locations.

 PFAS classes composition

Following ΣPFAS screening, PFAS class composition was evaluated. Detected analytes were grouped into major structural classes, including PFSAs, PFCAs, FT-derived acids/sulfonates, ECF-related sulfonamides, and modern replacement PFAS (Supplementary Figure S3). Class definitions and their analytical relevance were established based on published associations reported in the literature and were incorporated as part of the forensic framework. Class-level proportions were visualized using scaled pie charts, with color gradations distinguishing PFAS classes.

 Chain-length distribution

Chain-length distributions within PFCA and PFSA families were evaluated to characterize within-class composition. For each sampling location, homologs were grouped into short-chain species (≤C6 PFCAs and ≤C5 PFSAs) and long-chain homologs, following classifications used in prior studies (Leeson et al., 2021; Mejia-Avendaño et al., 2020; McGarr et al., 2023). Relative enrichment in these homolog groups was evaluated within each sample and compared across spatial and temporal datasets. Carbon-number patterns were visualized using scaled pie charts, with color shading indicating chain-length groupings.

 Linear vs. branched isomers

Isomer-resolved PFAS data were evaluated to differentiate ECF- and FT-associated production chemistries based on established linear and branched isomer distributions. Linear and branched isomers were distinguished chromatographically based on retention time separation using LC/MS/MS (Supplementary Figure S3). Branched isomers were quantified as a bulk group using integrated peak areas rather than as individual isomers. Linear-to-branched (L:Br) isomer ratios were calculated from the ratio of the linear isomer peak area to the summed peak area of all resolved branched isomers where sufficient signal was present. Isomer-resolved evaluation was restricted to perfluorooctanesulfonic acid (PFOS), for which branched isomers were consistently detected and quantifiable across the dataset (Supplementary Figure S5). Branched isomers of perfluorohexanesulfonic acid (PFHxS) and perfluorooctanoic acid (PFOA) were infrequently detected and, when present, occurred at concentrations near or below reliable quantification limits. In addition, chromatographic separation of linear and branched isomers for these compounds was poorer than that achieved for PFOS, with residual overlap between linear and branched signals further limiting confidence in isomer-specific quantification. As a result, L:Br ratios for PFHxS and PFOA were not evaluated to avoid overinterpretation of sparse, low-signal, and partially overlapping isomer data.

 Precursor dynamics

Precursor dynamics were evaluated by classifying detected PFAS precursors according to ECF- and FT-associated chemistries and assessing homolog distributions and expected terminal PFCA or PFSA products reported for precursor-derived transformation pathways (Liu et al., 2024; Ruyle et al., 2021a; Dasu et al., 2012).

 Tier 3 – Diagnostic ratios Chain-length–normalized ratio (ΣPFAS[≤C6]/ΣPFAS)

Chain-length-normalized ratios were calculated to quantify the proportion of short-chain PFAS (≤C6) relative to total PFAS. This metric was used to characterize the relative contribution of short-chain species within each sample and was evaluated in conjunction with PFSA/PFCA composition, sulfonate ratios, and isomer patterns as part of the integrated fingerprinting framework (Yao et al., 2018).

 PFSA-PFCA ratio (PFOS/PFOA)

PFOS/PFOA ratios were calculated for samples with quantified detections of both analytes. The ratio was used as a class-level compositional metric to compare relative sulfonate and carboxylate contributions and to evaluate consistency with reported PFCA-enriched and PFSA-enriched mixture profiles, including contrasts associated with fluoropolymer processing, wastewater-influenced inputs, and legacy ECF-derived materials (Prevedouros et al., 2006; Liu et al., 2024).

 PFSA-PFSA ratio (PFOS/PFHxS)

The PFOS/PFHxS ratio was calculated for samples with quantified detections of both analytes. Ratios were derived from measured concentrations and used as a normalized compositional metric to evaluate relative PFSA composition and to assess consistency with production-era signatures reported for ECF-based formulations in the peer-reviewed literature (Charbonnet et al., 2021; Reinikainen et al., 2022).

 Tier 4 – Spatial pattern analysis

Spatial PFAS distribution patterns were evaluated relative to the hydrostratigraphy and groundwater flow framework defined in the CSM. Variations in PFAS class composition and chain-length distributions were assessed along inferred groundwater flow paths across the site.

 Tier 5 – Temporal trends

Temporal analyses were conducted to evaluate changes in PFAS concentrations and composition across sampling events. Both concentration trends and relative-abundance metrics were evaluated across sampling rounds. Temporal comparisons incorporated class-level and compositional indicators, including PFSA/PFCA structure, sulfonate ratios, chain-length–normalized ratios, and precursor-related metrics.

For clarity, results from Tier 4 (spatial patterns) and Tier 5 (temporal trends) are presented together in Supplementary Section S3, as spatial organization and temporal evolution of PFAS mixtures are closely coupled within the site hydrogeologic framework. Integrated evaluation of these tiers enables assessment of how PFAS concentrations and composition vary along groundwater flow paths and evolve between sampling events. To support compound-level resolution within this combined spatial-temporal context, a PFAS fingerprinting matrix was also applied (Supplementary Figure S3). The matrix organizes individual PFAS by carbon number and structural subclass, facilitating visualization of concurrent changes in precursors and terminal perfluoroalkyl acids (PFAAs). Additional details on matrix construction and implementation are provided in Supplementary Section S3.

 Tier 6 – Multivariate clustering and ordination

Multivariate pattern-recognition methods were applied to evaluate similarity and dissimilarity among PFAS compositional profiles across the monitoring network. Unsupervised clustering and ordination techniques were used to analyze relative-abundance-normalized PFAS data and to assess covariance among analytes across samples (Xiao et al., 2012). Methods included cluster analysis and ordination approaches commonly used for multivariate environmental datasets (Joseph et al., 2023). Unsupervised clustering was performed on normalized compound-level PFAS profiles without imposing a priori source classifications. Partitioning Around Medoids (PAM) clustering was selected to identify representative medoids and to reduce sensitivity to outliers relative to centroid-based methods. Cluster solutions were evaluated over a range of cluster numbers (k), and internal validation metrics, including silhouette analysis and gap statistics, were used to guide selection of an appropriate solution.

Ordination methods were used as a complementary approach to visualize multivariate dissimilarity among samples and to evaluate gradients in PFAS compositional structure. Spatial context was incorporated post hoc to examine correspondence between multivariate groupings and site hydrogeologic features (Nguyen et al., 2025).

Statistical distinctness among clusters was evaluated using permutational multivariate analysis of variance (PERMANOVA). Additional details are provided in Supplementary Section S2.

 Results and discussion

The application of target-PFAS forensics framework is presented to illustrate its use within a structured analytical context.

Tier 1 – PFAS CSM data gathering

PFAS transport is represented within a shallow, high-permeability unconfined aquifer with predominantly south-southeast groundwater flow. Laterally continuous hydrostratigraphy provides hydraulic connectivity between source domains, downgradient monitoring locations, and adjacent surface-water bodies. Groundwater conditions are neutral to slightly alkaline and oxidizing, which favor sulfonate persistence and enhance mobility of short-chain PFAS. Potential receptors identified within the CSM include downgradient groundwater resources and connected surface-water features hydraulically linked to the shallow aquifer.

 Tier 2 – PFAS compositional and concentration screening Total PFAS concentration

Total PFAS concentrations (ΣPFAS), evaluated across all sampling events, show clear spatial separation among site domains (Figure 1). The lowest ΣPFAS values occur in wells located north of operational areas, while progressively higher concentrations are observed in the fire area, industrial corridor, and WWTP area. When compared to published Minnesota groundwater background concentrations, ΣPFAS values from northern locations fall within or below reported background ranges. In contrast, ΣPFAS concentrations measured in the fire area, industrial corridor, and WWTP area exceed background levels, with the magnitude of exceedance increasing from the fire area to the WWTP. Across the aggregated dataset, the fire area exhibits elevated ΣPFAS relative to background but lower concentrations than the industrial corridor. The industrial corridor shows consistently higher ΣPFAS, while the WWTP area displays the highest concentrations and the broadest concentration range among all site domains. These distributions indicate that groundwater PFAS concentrations at the site extend well beyond regional background levels within operational and downgradient areas.

Box plot graphic comparing total PFAS concentrations in nanograms per liter across four areas: Upgradient, Fire Area, Industrial Area, and WWTP Area, with an inset showing lower upgradient concentrations relative to Minnesota groundwater PFAS background levels marked by a red dashed line at 170 nanograms per liter.

Box plot graphic comparing total PFAS concentrations in nanograms per liter across four areas: Upgradient, Fire Area, Industrial Area, and WWTP Area, with an inset showing lower upgradient concentrations relative to Minnesota groundwater PFAS background levels marked by a red dashed line at 170 nanograms per liter. PFAS class composition

Wells within the former fire area contained a broad range of PFAS classes, including PFSAs, fluorinated sulfonamides (FASAs/FASAAs), fluorinated sulfonamidoethanols (FASEs), and PFCAs (Figure 2). PFSA-dominated profiles were observed near the fire zone, transitioning to mixed PFSA–PFCA compositions downgradient. The industrial corridor exhibited the most diverse class distribution in 2024, including PFSAs, PFCAs, sulfonamide-based precursors, and fluorotelomer sulfonates (FTSs), with an increased relative contribution of PFCAs. Samples near the onsite WWTP were PFCA-dominated, with low but consistent detections of PFSAs, FTSs, FASAs, and FASAAs.

Side-by-side pie chart maps display detected PFAS classes at multiple PFAS sampling locations in 2018 and 2024. Pie size shows total PFAS concentration, while segments indicate PFAS class.

Side-by-side pie chart maps display detected PFAS classes at multiple PFAS sampling locations in 2018 and 2024. Pie size shows total PFAS concentration, while segments indicate PFAS class. Chain-length distribution

Groundwater within the former fire area displayed a carbon-number distribution dominated by long-chain PFSAs, primarily PFOS and PFHxS with lower relative contributions from long-chain PFCAs and sulfonamide precursors (perfluorooctanesulfonamide [FOSA], n-methyl perfluorooctanesulfonamide [NMeFOSA], n-ethyl perfluorooctanesulfonamide [NEtFOSA], n-methyl perfluorooctanesulfonamidoacetic acid [NMeFOSAA], n-ethyl perfluorooctanesulfonamidoacetic acid [NEtFOSAA]). The industrial corridor exhibited broader carbon-number distributions characterized by substantial contributions from fluorotelomer sulfonates (8:2 and 6:2 FTS) and PFCAs spanning both long- and short-chain homologs (Figure 3). Samples associated with the onsite WWTP showed the strongest short-chain enrichment across the site, dominated by perfluorohexanoic acid (PFHxA), perfluoropentanoic acid (PFPeA), and perfluorobutanoic acid (PFBA), with additional contributions from PFOA and short-chain fluorotelomer sulfonates (4:2 and 6:2 FTS).

Side-by-side graphic shows groundwater PFAS concentrations at an industrial site for 2018 and 2024, with colors assigned to distinguish compound classes and chain length (darker tones for higher-molecular-weight species and lighter tones for more mobile, lower-molecular-weight compounds).

Side-by-side graphic shows groundwater PFAS concentrations at an industrial site for 2018 and 2024, with colors assigned to distinguish compound classes and chain length (darker tones for higher-molecular-weight species and lighter tones for more mobile, lower-molecular-weight compounds). Linear vs. branched isomers

L:Br PFOS ratios measured in 2024 show limited spatial variability across the site (Supplementary Figure S6). Elevated ratios are confined to isolated locations within the former fire-area domain, while values in the industrial corridor and WWTP zones fall within a narrower and lower range. Overall, PFOS isomer distributions in 2024 exhibit reduced linear enrichment and increased spatial uniformity across site domains.

 Precursor dynamics

Spatial patterns in precursor and terminal PFAS distributions observed in 2024 show clear structure across the site (Figures 4, 5). ECF-derived sulfonamide precursors (FOSA, NMeFOSA, NEtFOSA, NMeFOSAA, NEtFOSAA) were most prominent in the former fire-area domain, with lower-level occurrences extending into the industrial corridor and WWTP areas. Fluorotelomer sulfonates (6:2 and 8:2 FTS) were spatially concentrated in the industrial corridor and WWTP zones, where they co-occur with elevated terminal PFCA concentrations. Terminal PFCAs in 2024 exhibit their highest concentrations within and downgradient of the WWTP zone, while PFSAs remain spatially focused in the fire-area domain. This spatial separation between PFSA-dominated and PFCA-dominated regions is evident across the interpolated surfaces.

Spatial interpolation maps of PFAS precursor classes in groundwater samples collected in 2018 and 2024. (A,B) ECF-derived sulfonamide precursors, and (C,D) fluorotelomer sulfonates (FTSs). Color gradients represent molar concentration (µmol/L), with darker shades indicating higher levels. Dashed boxes mark the key source domains.

Spatial interpolation maps of PFAS precursor classes in groundwater samples collected in 2018 and 2024. (A,B) ECF-derived sulfonamide precursors, and (C,D) fluorotelomer sulfonates (FTSs). Color gradients represent molar concentration (µmol/L), with darker shades indicating higher levels. Dashed boxes mark the key source domains.

Spatial interpolation of terminal PFAS classes in groundwater samples collected in 2018 and 2024. (A,B) PFCAs, and (C,D) PFSAs. Color gradients represent molar concentration (µmol/L), with darker shades indicating higher levels. Dashed boxes mark the key source domains.

Spatial interpolation of terminal PFAS classes in groundwater samples collected in 2018 and 2024. (A,B) PFCAs, and (C,D) PFSAs. Color gradients represent molar concentration (µmol/L), with darker shades indicating higher levels. Dashed boxes mark the key source domains. Tier 3 – Diagnostic ratios ≤C6 PFCA ratio

Elevated ≤ C6 PFCA ratios observed in 2024 are spatially concentrated within the industrial corridor and the WWTP domain, identifying these areas as the primary zones of short-chain PFCA dominance (Supplementary Figure S7A). The spatial footprint of elevated ratios extends across much of the industrial-wastewater domain, while values elsewhere remain comparatively lower. Higher ≤ C6 PFCA ratios are also observed at select upgradient locations; however, these values coincide with very low total PFAS concentrations and compositions dominated by short-chain PFCAs. Consequently, ratio magnitudes at these locations reflect sensitivity to low absolute concentrations rather than elevated short-chain PFCA mass. Across the site, interpretation of the ≤C6 PFCA ratio should be interpretated in conjunction with total PFAS concentrations and class- and chain-length distributions.

 PFOS/PFOA ratio

PFOS/PFOA ratios measured in 2024 exhibit spatial variability across the site, with lower ratios primarily observed within the industrial corridor and WWTP domains (Supplementary Figure S7B). Several locations are not represented in the ratio maps due to non-detections of PFOA, reflecting limitations of ratio-based metrics under spatially heterogeneous detections. Where PFOS/PFOA ratios could be calculated, lower ratios co-occur with site areas showing higher relative PFCA contributions in class-level analyses. Accordingly, PFOS/PFOA ratios should be interpreted together with absolute concentrations and PFAS class composition.

 PFOS/PFHxS ratio

PFOS/PFHxS ratios in 2024 are low across most of the site, with elevated values restricted to a small number of locations (Supplementary Figure S7C). Higher ratios occur primarily within the former fire-area domain and at immediately downgradient wells, where PFOS constitutes a greater relative fraction of PFSAs than PFHxS. A single location within the industrial corridor exhibits a markedly elevated PFOS/PFHxS ratio. This value reflects very low PFHxS detections combined with low PFOS concentrations that remain minor relative to total PFAS at that location, indicating a denominator-driven effect rather than PFOS dominance within the overall PFAS mixture. At locations where the ratio is calculable, PFOS/PFHxS values reflect localized variability in PFSA composition rather than a site-wide pattern.

 Tier 4-5 – Spatial and temporal trends

PFAS concentrations and compositions exhibit persistent spatial organization aligned with the south-southeast groundwater flow direction across both sampling events. Total PFAS magnitudes increase systematically from the former fire area through the industrial corridor and reach maximum values in the onsite WWTP domain. Between 2018 and 2024, spatial patterns remain coherent while concentration magnitudes and mixture composition evolve (Figures 2, 3). FTSs show substantial declines across the site between sampling events, most prominently within the industrial corridor and WWTP domains (Figure 4). In contrast, ECF-associated compounds display more heterogeneous temporal behavior (Figure 4). PFSA concentrations within the former fire-area domain remain elevated and broadly similar in 2018 and 2024, whereas lower-level PFSA detections observed in portions of the industrial corridor and WWTP in 2018 remain detectable in 2024 (Figure 5). Coincident with the decline in FTSs, total PFCA mass increases markedly across the site, rising from approximately 40,000 μmol/L in 2018 to approximately 100,000 μmol/L in 2024, with the largest increases occurring within and downgradient of the industrial corridor and WWTP domains.

To identify which individual PFCAs contribute to the increase in total PFCA mass, the PFAS fingerprinting matrix was evaluated for a representative monitoring well in the industrial corridor (Figure 6). Compound-level patterns show concurrent increases in PFBA, PFPeA, PFHxA, perfluoroheptanoic acid (PFHpA), and PFOA, coincident with declining fluorotelomer sulfonates. Both short- and longer-chain PFCAs contribute to the net PFCA increase at this location, potentially indicating that the observed mass gain is distributed across multiple precursor sources rather than dominated by a single fluorotelomer compound.

PFAS profile changes at monitoring location 22X (2018 vs. 2024). A marked decline in 6:2 FTS coincides with significant increases in multiple PFCAs (PFBA, PFPeA, PFHxA, PFHpA, PFOA), indicating ongoing biotransformation.

PFAS profile changes at monitoring location 22X (2018 vs. 2024). A marked decline in 6:2 FTS coincides with significant increases in multiple PFCAs (PFBA, PFPeA, PFHxA, PFHpA, PFOA), indicating ongoing biotransformation. Tier 6 – Multivariate clustering

Multivariate clustering resolved four compositionally distinct PFAS mixture groups across the monitoring network (Figure 7). The four-cluster solution explains a substantial fraction of the multivariate variability in PFAS composition (PERMANOVA R2 = 0.30, p < 0.01), with moderate separation among clusters (average silhouette width = 0.52).

Unsupervised Partitioning Around Medoids (PAM) clustering of groundwater samples based on normalized PFAS profiles, revealing four distinct compositional clusters. Left panel: ordination plot illustrating cluster separation and internal cohesion. Right panel: spatial distribution of clusters across the site in 2018, highlighting associations with key source domains—industrial corridor (Cluster 1, green), fire-event area (Cluster 2, orange), WWTP influence zone (Cluster 3, purple), and background conditions (Cluster 4, pink). This analysis uses a rank-based Euclidean distance clustering framework to support exploratory evaluation of compositional structure across the monitoring network.

Unsupervised Partitioning Around Medoids (PAM) clustering of groundwater samples based on normalized PFAS profiles, revealing four distinct compositional clusters. Left panel: ordination plot illustrating cluster separation and internal cohesion. Right panel: spatial distribution of clusters across the site in 2018, highlighting associations with key source domains—industrial corridor (Cluster 1, green), fire-event area (Cluster 2, orange), WWTP influence zone (Cluster 3, purple), and background conditions (Cluster 4, pink). This analysis uses a rank-based Euclidean distance clustering framework to support exploratory evaluation of compositional structure across the monitoring network.

The resulting clusters exhibit strong spatial coherence and align closely with operational domains identified through independent concentration, compositional, and geospatial analyses. Cluster 1, spatially associated with the industrial corridor, is characterized by elevated relative contributions of fluorotelomer sulfonates and associated PFCAs, reflecting a telomer-influenced industrial mixture structure. Cluster 2 corresponds to the former fire-area domain and is enriched in PFSAs, consistent with a sulfonate-dominated mixture distinct from other site areas. Cluster 3 encompasses wells within and downgradient of the onsite WWTP and is defined by pronounced enrichment in short-chain PFCAs. This cluster shows partial compositional overlap with Cluster 1, consistent with hydraulic connectivity and wastewater-mediated transfer of PFAS mass from the industrial corridor. Cluster 4 contains wells with low total PFAS concentrations and no dominant compound class, representing background or minimally influenced groundwater conditions.

 Discussion and conclusion

Characterizing PFAS sources and plume evolution at complex sites remains challenging due to the large number of analytes, overlapping production eras, precursor-product relationships, and strong coupling between transport and transformation processes. Traditional single-compound or single-ratio approaches are often insufficient to resolve these complexities, particularly where multiple historical inputs coexist and PFAS composition evolves over time. The tiered PFAS fingerprinting framework presented here addresses these challenges by integrating concentration magnitude, compositional structure, compound-level behavior, and multivariate patterns into a coherent interpretive workflow. A key strength of this framework is its flexibility and scalability. Individual tiers can be applied independently when data are limited, while the full framework enables increasingly refined interpretation as analytical resolution improves. By combining bulk concentration screening, class- and chain-length structure, diagnostic ratios, compound-specific matrix analysis, and multivariate clustering, the approach reduces reliance on any single diagnostic indicator and instead emphasizes convergence among multiple lines of evidence.

Application of the framework across the evaluated domains illustrates its potential utility in resolving domain-scale differences in PFAS mixtures and their spatial-temporal evolution. Across analytical tiers, PFAS distributions consistently differentiate the former fire area, industrial corridor, and WWTP domains, with signatures that persist spatially while exhibiting varying degrees of temporal modification. These patterns suggest that historical inputs continue to influence present-day groundwater composition, with superimposed changes driven by transport, precursor turnover, and differential mobility. Compound-level evaluation using the PFAS fingerprinting matrix provides additional resolution by identifying which individual PFCAs contribute to observed increases in total PFCA mass. The matrix highlights concurrent increases across multiple PFCA homologs, indicating that PFCA enrichment is not dominated by a single compound. However, these observations do not uniquely constrain the number or identity of contributing precursors, and alternative explanations, including mixed source inputs or unquantified precursor classes, cannot be excluded. Diagnostic ratios and isomer patterns provide supporting context within the framework but also demonstrate inherent limitations. Ratio-based metrics are sensitive to detection frequency, concentration magnitude, and analytical reporting limits, particularly in heterogeneous datasets. Accordingly, these indicators are interpreted cautiously and only in conjunction with absolute concentrations, compositional patterns, and multivariate structure. Multivariate clustering resolves compositionally distinct PFAS mixture groupings that are spatially coherent and broadly aligned with operational domains and inferred hydrologic connectivity. These groupings corroborate patterns identified through independent concentration-, composition-, and ratio-based analyses.

Collectively, the spatial and temporal trends observed within this application suggest persistent domain-specific PFAS signatures with superimposed compositional modification along groundwater flow paths. While the available data do not allow definitive attribution of individual transformation mechanisms at all locations, the convergence of multiple analytical tiers supports interpretation of PFAS plume evolution driven by a combination of historical inputs, precursor-related processes, and differential mobility.

 Data availability statement

The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

 Author contributions

JZ: Conceptualization, Formal Analysis, Methodology, Writing – original draft, Writing – review and editing. FP: Writing – original draft, Writing – review and editing. AF: Formal Analysis, Methodology, Writing – review and editing. S-YC: Conceptualization, Methodology, Writing – review and editing.

 Acknowledgements

The authors acknowledge the use of PFluorensics™, the PFAS fingerprinting and visualization framework developed at Jacobs, for data processing, compositional analysis, geospatial visualization, and clustering used in this study. All analytical algorithms and parameter settings used in this study are fully described in the Methods and Supporting Information to enable independent reproduction of the results. No proprietary algorithms were required to generate the findings reported herein. The authors also thank internal Jacobs collaborators who contributed to development, testing, and refinement of the PFluorensics™ toolset and provided technical insight during analysis and interpretation.

 Conflict of interest

Author JZ was employed by Jacobs.

Author FP was employed by Jacobs.

Authors AF and S-YC were employed by Jacobs.

 Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

 Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

 Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fenve.2026.1766573/full#supplementary-material

References Annunziato K. M. Doherty J. Lee J. Clark J. M. Liang W. Clark C. W. (2020). Chemical characterization of a legacy aqueous film-forming foam sample and developmental toxicity in zebrafish (Danio Rerio). Environ. Health Perspect. 128 (9), 097006. 10.1289/EHP6470 32966100 Benotti M. J. Fernandez L. A. Peaslee G. F. Douglas G. S. Uhler A. D. Emsbo-Mattingly S. (2020). A forensic approach for distinguishing PFAS materials. Environ. Forensics 21 (3–4), 319–333. 10.1080/15275922.2020.1771631 Charbonnet J. A. Rodowa A. E. Joseph N. T. Guelfo J. L. Field J. A. Jones G. D. (2021). Environmental source tracking of Per- and Polyfluoroalkyl substances within a forensic context: current and future techniques. Environ. Sci. Technol. 55 (11), 7237–7245. 10.1021/acs.est.0c08506 33983714 Dasu K. Liu J. Lee L. S. (2012). Aerobic soil biodegradation of 8:2 fluorotelomer stearate monoester. Environ. Sci. Technol. 46 (7), 3831–3836. 10.1021/es203978g 22372635 Eriksson U. Haglund P. Kärrman A. (2017). Contribution of precursor compounds to the release of Per- and Polyfluoroalkyl substances (PFASs) from waste water treatment plants (WWTPs). J. Environ. Sci. 61 (November), 80–90. 10.1016/j.jes.2017.05.004 29191318 He A. Liang Y. Li F. Lu Y. Liu C. Li J. (2022). Vital environmental sources for multitudinous fluorinated chemicals: new evidence from industrial byproducts in multienvironmental matrices in a fluorochemical manufactory. Environ. Sci. Technol. 56 (23), 16789–16800. 10.1021/acs.est.2c04372 36354080 Joerss H. Menger F. (2023). The complex ‘PFAS World’ - how recent discoveries and novel screening tools reinforce existing concerns. Curr. Opin. Green Sustain. Chem. 40 (April), 100775. 10.1016/j.cogsc.2023.100775 Joseph N. T. Schwichtenberg T. Cao D. Jones G. D. Rodowa A. E. Barlaz M. A. (2023). Target and suspect screening integrated with machine learning to discover Per- and Polyfluoroalkyl substance source fingerprints. Environ. Sci. Technol. 57 (38), 14351–14362. 10.1021/acs.est.3c03770 37696050 Joseph N. T. Droz B. Schwichtenberg T. Oetjen K. Sühnholz S. Jones G. D. (2025). Discovery of comprehensive sets of chemical constituents as markers of PFAS sources through a nontarget screening and machine learning approach. Environ. Sci. Technol. 59 (42), 22852–22865. 10.1021/acs.est.5c07560 41101933 Langberg H. A. Hale S. E. Breedveld G. D. Jenssen B. M. Jartun M. (2022). A review of PFAS fingerprints in fish from Norwegian freshwater bodies subject to different source inputs. Environ. Sci. Process. Impacts 24 (2), 330–342. 10.1039/D1EM00408E 35079763 Leeson A. Thompson T. Stroo H. F. Anderson R. H. Speicher J. Mills M. A. (2021). Identifying and managing aqueous film-forming foam-derived PFAS in the environment. Environ. Toxicol. Chem. 40 (1), 24–36. 10.1002/etc.4894 33026660 Liu Y. D’Agostino L. A. Qu G. Jiang G. Martin J. W. (2019). High-resolution mass spectrometry (HRMS) methods for nontarget discovery and characterization of Poly- and per-Fluoroalkyl substances (PFASs) in environmental and human samples. TrAC Trends Anal. Chem. 121 (December), 115420. 10.1016/j.trac.2019.02.021 Liu M. Glover C. M. Munoz G. Duy S. V. Sauvé S. Liu J. (2024). Hunting the missing fluorine in aqueous film-forming foams containing Per- and Polyfluoroalkyl substances. J. Hazard. Mater. 464, 133006. 10.1016/j.jhazmat.2023.133006 37988941 Malik P. Nandini D. Tripathi B. P. (2024). Firefighting aqueous film forming foam composition, properties and toxicity: a review. Environ. Chem. Lett. 22 (4), 2013–2033. 10.1007/s10311-024-01739-x McGarr J. T. Mbonimpa E. G. McAvoy D. C. Soltanian M. R. (2023). Fate and transport of Per- and Polyfluoroalkyl substances (PFAS) at aqueous film forming foam (AFFF) discharge sites: a review. Soil Syst. 7 (2), 53. 10.3390/soilsystems7020053 Mejia-Avendaño S. Zhi Y. Yan B. Liu J. (2020). Sorption of Polyfluoroalkyl surfactants on surface soils: effect of molecular structures, soil properties, and solution chemistry. Environ. Sci. Technol. 54 (3), 1513–1521. 10.1021/acs.est.9b04989 31922402 MPCA (2024). Minnesota pollution control agency - PFAS ambient background concentrations. PFAS Studies and Reports. Available online at: https://www.pca.state.mn.us/sites/default/files/tdr-g1-25.pdf. (Accessed August 12, 2025) Nguyen D. Cadle S. Verstraet T. Kammer L. Van Scoy T. Anding M. (2025). Novel insights from multivariate analysis of metadata from a large PFAS remedial investigation dataset. J. Hazard. Mater. 500 (December), 140539. 10.1016/j.jhazmat.2025.140539 41270413 Prevedouros K. Cousins I. T. Buck R. C. Korzeniowski S. H. (2006). Sources, fate and transport of perfluorocarboxylates. Environ. Sci. Technol. 40 (1), 32–44. 10.1021/es0512475 16433330 Reinikainen J. Perkola N. Äystö L. Sorvari J. (2022). The occurrence, distribution, and risks of PFAS at AFFF-impacted sites in Finland. Sci. Total Environ. 829 (July), 154237. 10.1016/j.scitotenv.2022.154237 35292316 Ruyle B. J. Thackray C. P. McCord J. P. Strynar M. J. Mauge-Lewis K. A. Fenton S. E. (2021a). Reconstructing the composition of Per- and Polyfluoroalkyl substances in contemporary aqueous film-forming foams. Environ. Sci. Technol. Lett. 8 (1), 59–65. 10.1021/acs.estlett.0c00798 33628855 Ruyle B. J. Pickard H. M. LeBlanc D. R. Tokranov A. K. Thackray C. P. Hu X. C. (2021b). Isolating the AFFF signature in coastal watersheds using oxidizable PFAS precursors and unexplained organofluorine. Environ. Sci. Technol. 55 (6), 3686–3695. 10.1021/acs.est.0c07296 33667081 Strynar M. McCord J. Newton S. Washington J. Barzen-Hanson K. Trier X. (2023). Practical application guide for the discovery of novel PFAS in environmental samples using high resolution mass spectrometry. J. Expo. Sci. Environ. Epidemiol. 33 (4), 575–588. 10.1038/s41370-023-00578-2 37516787 U.S. EPA (2024). Method 1633, Revision A Analysis of Per- and Polyfluoroalkyl Substances (PFAS) in Aqueous, Solid, Biosolids, and Tissue Samples by LC-MS/MS. Available online at: https://www.epa.gov/system/files/documents/2024-12/method-1633a-december-5-2024-508-compliant.pdf. (Accessed August 12, 2025). Xiao F. Halbach T. R. Simcik M. F. Gulliver J. S. (2012). Input characterization of Perfluoroalkyl substances in wastewater treatment plants: source discrimination by exploratory data analysis. Water Res. 46 (9), 3101–3109. 10.1016/j.watres.2012.03.027 22483712 Yao Y. Burgess J. Volchek K. Brown C. E. (2018). “Short-chain PFAS: their sources, properties, toxicity, environmental fate, and treatment,” in Perfluoroalkyl substances in the environment (Boca Raton: CRC Press).

Edited by: Hui Lin, Dongguan University of Technology, China

Reviewed by: Caiming Tang, Dongguan University of Technology, China

Bei Zhang, Zhejiang University, China