Purpose and Scope

This report presents the results of groundwater sampling and analysis to describe the detection and concentrations of PFAS in groundwater from 23 previously sampled wells in the GW-BVA. To understand the comparability of PFAS analytical results generated through different adaptations of the EPA 537.1 method, analytical results of groundwater samples collected by this study and analyzed at a contract laboratory using one proprietary adaptation (method 1) were compared with results from samples that were in most cases sequentially collected from the same wells and analyzed at a different contract laboratory by a slightly different proprietary adaptation (method 2).

Concentrations of PFAS in groundwater were compared with Ohio action levels and Federal health-risk-based guidance from the EPA. Groundwater PFAS data were also compared with several types of land use, chemical and hydrologic data, and literature on the development and common use of PFAS to understand compound detections relative to likely PFAS use and factors that affect groundwater vulnerability to contamination. Tritium-helium-3 and tritium-based initial estimates of groundwater age from prior groundwater analyses from wells sampled by this study (Hinkle and others, 2010), and tritium-based groundwater-age categories from the coordinated NWQP sampling (McMahon and others, 2022a) were compared with PFAS detections in groundwater and to a literature-based classification of the likely history of common development, use, and potential environmental release of PFAS. Redox or reduction/oxidation categories of prior groundwater samples (1999–2000) and samples collected by this study from the same wells were classified (Jurgens and others, 2009) to understand how redox-related processes may relate to PFAS detections in groundwater and the transformation of PFAS precursor compounds into terminal degradation products (Interstate Technology and Regulatory Council, 2022).

Background Information About Per- and Polyfluoroalkyl Substances

PFAS are a class of synthetic organic compounds with widespread uses in industrial processes and consumer products since the 1940s (Interstate Technology and Regulatory Council, 2022). For comprehensive discussion, references, summary of naming conventions, use, properties, fate and transport, environmental detections, and health effects of PFAS and other topics relevant to sampling, treatment, and regulation of PFAS, the reader is referred to technical resources given by the Interstate Technology and Regulatory Council (2022). The rest of this section summarizes details from that and other references.

PFAS produced by electrochemical fluorination that contained perfluorooctanoate (PFOA) or perfluorooctanesulfonate (PFOS) came into production in the late 1940s (Buck and others, 2011, p. 520–521; Prevedouros and others, 2006, p. 32–33). Products with PFOA and PFOS came into common use in the early to mid-1950s until 1972 in stain and water-resistant products (California Department of Toxic Substances Control, 2019, p. 15; Prevedouros and others, 2006, p. 32–33) and from about 1964 to 1971 in different firefighting foam preparations (Prevedouros and others, 2006; Gipe and Peterson, 1972, p. 1–2; Sheinson and others, 2002, p. 2; Dlugogorski and Schaefer, 2021, p. 7). PFAS have been used in consumer products, such as grease, oil, stain, heat resistant coatings for paper products and food packaging, textiles, leather, carpets, and non-stick cookware (Interstate Technology and Regulatory Council, 2022). Industrial and commercial applications of PFAS have included photolithography, semiconductor manufacture, electrical wire insulation, fire suppression as aqueous film forming foam, and metal plating and etching for fume suppression, corrosion, and wear prevention, and in post-plating cleaning (Interstate Technology and Regulatory Council, 2022). Several PFAS, including PFOA and PFOS, have uses in the United States that have decreased since about 2000 (3M Company, 2000; EPA, 2000, 2017, 2022a; Seow, 2013). Fluorotelomers were introduced as firefighting foams in the 1970s and were more commonly used for that purpose after about 2001 (Seow, 2013). Release of PFAS into the environment has been widespread because of release during their production as chemicals, secondary release during manufacture, use and disposal of products that use PFAS, and tertiary release in waste streams from these processes, such as in residues from wastewater treatment and waste disposal (Interstate Technology and Regulatory Council, 2022).

PFAS are fluorinated aliphatic linear and branched carbon-chain organic compounds (Interstate Technology and Regulatory Council, 2022; Renner, 2006). They are termed fluorosurfactants because of their organofluorine-based chemistry and because of the combined hydrophobic (water repelling, hydrocarbon affinity) and hydrophilic (anionic or water affinity) features of their chemical structures (Renner, 2006). Fluorosurfactants have a hydrophobic functional group of electronegative fluorine atoms bonded to carbon in several lengths. Older fluorosurfactants such as PFOA or PFAS typically had longer aliphatic carbon chains with eight carbon atoms and more fluorine atoms as functional groups, making them more hydrophobic and more environmentally persistent. Newer classes of organofluorine compounds include the ammonium salt of hexafluoropropylene oxide dimer acid (HFPO–DA) fluoride that is used to produce fluoropolymers to replace phased-out PFOA. Several shorter aliphatic four-carbon-chain organofluorine surfactants, including perfluorobutanesulfonate (PFBS) and perfluorohexanesulfonate (PFHxS), have come into common use as replacements for PFOS-based applications (Bogdan, 2019; Renner, 2006; Buck and others, 2011; Minnesota Pollution Control Agency 2021). Before 2002, PFBS was also produced as a residual from electrochemical fluorination production of perfluorooctane sulfonyl fluoride and was in firefighting foam (Bogdan, 2019).

Human exposure to PFAS can happen through consuming water or food that contains PFAS contaminants or that was packaged in PFAS containing material and through consumer and occupational use of products that contain PFAS (Agency for Toxic Substance and Disease Registry, 2022). Results from analyses of specific PFAS in blood serum, sampled from the U.S. population since 1999, indicated wide-ranging exposure to several PFAS, including PFOS, PFOA, PFHxS, and perfluorononanoate (PFNA) (National Biomonitoring Program, 2017). Results from epidemiology studies indicate possible links between higher levels of several types of PFAS in human blood serum and changes to cholesterol levels, immune system function, and metabolism and increased risk of obesity and type-2 diabetes, non-alcoholic fatty liver disease, kidney cancer, and thyroid disease (Agency for Toxic Substance and Disease Registry, 2022; Kielsen and others, 2016; Liu and others, 2018; Qi and others, 2020; Shearer and others, 2021). Several PFAS, including PFOS, PFNA, perfluorodecanoate, and longer chain perfluoroalkyl carboxylic acids, such as PFOA, have been noted to biomagnify in fish and wildlife through the food chain (Burkhard, 2021).

Several classes of PFAS were analyzed from water samples in this study (table 1), and concentrations of specific PFAS from those analyses were compared with State guidance for drinking water and Federal health-risk-based guidance (table 2). The Ohio Environmental Protection Agency and Ohio Department of Health have established action levels for PFOA and PFOS concentrations in drinking water of 70 nanograms per liter (ng/L) for a single concentration of either PFOA or PFOS in a sample or the sum of PFOA and PFOS concentrations in a sample (Ohio Environmental Protection Agency and Ohio Department of Health, 2019; table 2). The Ohio action levels are consistent with the EPA health advisories established in 2016 for lifetime exposure to PFOA and PFOS (EPA, 2016a, 2016b, 2016c). Ohio action levels and Federal health-risk-based guidance also have been issued for PFBS, PFHxS, and PFNA (table 2).

Interim drinking water health advisories (IHA) for PFOS and PFOA and a health advisory for PFBS were established by the EPA in June 2022 and were used in this report to compare with PFAS analyses in groundwater (EPA, 2022a, 2022c, 2022d; table 2). The IHA guidances for PFOS and PFOA were established at very low concentrations of 0.02 ng/L and 0.004 ng/L, respectively, based on peer-reviewed data published after the prior EPA guidance issued in 2016 (EPA, 2016b, 2016c, 2022a). The IHA guidance is about 3,500 times less for PFOS and 17,500 times less for PFOA than the 2016 EPA health guidances and Ohio action levels as of 2019. The health advisory of 2,000 ng/L for PFBS established in June 2022 is slightly less than its 2019 Ohio action level of 2,100 ng/L (table 2). The IHA guidance for PFOS and PFOA were defined by the EPA to be protective for the most sensitive non-cancer effect identified in their research, decreased immunity, as defined by decreased serum antibody concentrations after vaccination in children (EPA, 2022a). The IHA guidances and their scientific basis are under review as of December 2022, and could change as a result (EPA, 2022a).

Health-based screening guidance for PFAS, referred to as “minimal risk level,” has been published by the Agency for Toxic Substance and Disease Registry to use in comparison with PFAS concentrations in drinking water as an indicator of whether potential public health effects may be evaluated further; this guidance is included in table 2 for reference (Agency for Toxic Substance and Disease Registry, 2021). The minimal risk levels were derived by applying dose-based minimal risk levels to assumed adult and small child body weights, plus assumed daily water intake rates.

Study Area

The study area includes parts of Butler, Champaign, Clark, Greene, Hamilton, Miami, Montgomery, Shelby, and Warren Counties within the Great Miami River, Little Miami River, and Whitewater River watersheds in southwestern Ohio (fig. 1). The study focused on groundwater from areas underlain or adjacent to the GW-BVA in those counties (fig. 1). The Great and Little Miami River watersheds drain about 5,880 square miles of southwestern Ohio (Rowe and others, 1997), and the Whitewater River watershed drains 145 square miles of southwestern Ohio and 1,329 square miles of southeast Indiana (Beaty and Clendenon, 1988). As of 2019, the estimated population of the counties in the study area included about 2.5 million residents (U.S. Census Bureau, 2020).

Most public, industrial, agricultural, and commercial water users in the study area have used groundwater from the GM-BVA as the principal or sole source of their water supply (Rowe and others, 1997). The GM-BVA and a larger area of adjacent alluvial and outwash deposits were classified as the Greater Miami sole-source aquifer (EPA, 1988a, 1988b; Ohio Environmental Protection Agency, 2009). In parts of the study area adjacent to the GM-BVA, drinking water is also produced from sand-and-gravel lenses in till and from underlying carbonate bedrock (Rowe and others, 1997; Debrewer and others, 2000).

Land use in the GM-BVA in the study area includes ranges of residential densities from low (130 to 999 people per square mile) to high (5,180 to 12,999 people per square mile), as characterized using 1990 population data (Debrewer and others, 2000, p. 45). Dominant land uses over the entire GM-BVA as of about 1990 were agricultural (69.4 percent) and urban (21.5 percent; Debrewer and others, 2000, p. 35). Urban land use in these areas included residential, commercial, and industrial classifications. Agricultural land use was described as chiefly devoted to the production of corn and soybeans (Debrewer and others, 2000). The remaining 9.9 percent of land use consisted of forested land, open water, wetlands, and small areas of mined or quarried land (Debrewer and others, 2000, p. 45).

By comparison, more recent data on dominant land uses as of 2011 in the Great Miami River watershed only, including parts of the watershed not underlain by the GM-BVA, were similarly classified to be agricultural (68.04 percent), urban (17.82 percent), and forested (11.54 precent). The remaining land in the Great Miami River watershed as of 2011 was covered by a mix of open water (1.01 percent), wetlands (0.26 percent) and other land uses (grassland, shrub/scrub vegetation, and barren/mined; 1.33 percent; Miami Conservancy District, 2015).

Hydrogeologic Framework

The hydrogeologic and environmental framework and generalized descriptions of flow directions and groundwater recharge summarized by Debrewer and others (2000) and Dumouchelle (1998) were used to describe the setting of sampled well locations within the GM-BVA. The geology, hydrogeology, water-level fluctuations, and water quality of the GM-BVA and land use and potential contaminant sources overlying the GM-BVA and adjacent areas have extensively been described by several authors, as summarized in Debrewer and others (2000). This report provides an overall description of the hydrogeologic framework, factors affecting groundwater recharge rates, flow and vulnerability to contaminants, and other factors in the study area to understand the groundwater quality results that are the focus of this work.

The GM-BVA principally consists of water deposited units of stratified and unstratified sand and gravels, interbedded with less permeable fine-grained glacial deposits that include clay-rich till, that fill a system of subsurface (buried) bedrock valleys (Dumouchelle, 1998). The bedrock valleys are incised into Ordovician period shale units and Devonian and Silurian period carbonate bedrock units (Debrewer and others, 2000; Sheets, 2007, fig. 2). The GM-BVA is adjacent to upland glacial deposits consisting of clay rich till that include less extensive interbedded sand and gravels that were deposited over Devonian and Silurian carbonate bedrock units (fig. 2). Sediments that comprise the GM-BVA range from 0 to nearly 400 feet (ft) in thickness and are commonly 150 to 200 ft in thickness (Norris and Spieker, 1966; Debrewer and others, 2000).

The permeable sand and gravel deposits of the GM-BVA consist of post-glacial unconsolidated alluvium, alluvial terrace deposits, glacial outwash, and adjacent permeable kame-derived sands from Wisconsinan glaciation (Pavey and others, 1999; Dumouchelle, 1998). Principal glacial-derived aquifer deposits within the GM-BVA include well-sorted and stratified sand and gravel deposits mapped at lower altitudes as valley-train outwash, outwash terraces at intermediate altitudes, and as undifferentiated outwash deposits at higher altitudes (terraces and low plains). At the outwash deposit margins, the GM-BVA includes kames and kame terraces that consist of poorly sorted and bedded sand and gravel (Pavey and others, 1999). Water-bearing units in adjacent upland areas that border the GM-BVA consist of unconsolidated fine-grained tills, sands, and gravels derived from Illinoian and Wisconsinan glaciation. Underlying and adjacent bedrock units were not represented in sampling reported by this study.

Groundwater in the GM-BVA regionally flows along the topographic gradient (Debrewer and others, 2000). Local groundwater flow is directed generally from recharge in nearby upland areas and along valley walls to where it discharges to streams through base flow or to discharges at major supply wells, as depicted by Dumouchelle (1998, plate 1; Sheets, 2007, p. 7–8). The complex distribution of sediments and their hydrogeologic properties within the GM-BVA, when combined with variations in hydraulic gradient and distributions of groundwater production, and hydraulic controls along major streams make it difficult to characterize local flow directions except in areas where all these factors are well characterized.

The GM-BVA is vulnerable to modern, post-1952 contamination, as indicated through recharge, hydraulic conductivity, and other aquifer characteristics. Typical pathways of groundwater in the GM-BVA from a recharge area to discharge at surface water or to a withdrawal well include vertical and horizontal flow components (Sheets, 2007, p. 7–8). Parts of the GM-BVA, with coarse-grained sediments at land surface, have higher recharge rates of about 6 to 15 inches per year relative to areas with fine-grained tills at land surface having smaller rates of less than 3 to about 5 inches per year (Norris and Spieker, 1966; Dumouchelle and others, 1993; Dumouchelle, 1998). The parts of an aquifer with higher recharge rates are overall more vulnerable to potential contaminants entering groundwater than those with smaller recharge rates. Recharge to groundwater in parts of the GM-BVA is locally induced through streambed infiltration from adjacent streams where supply-well pumping reverses normal patterns of groundwater flow and discharge to those streams (Dumouchelle, 1998). Such pumping also locally intercepts groundwater that would normally discharge to those streams.

Aquifer-test results and well yields in the GM-BVA also support the aquifer being characterized as very permeable and productive with rapid flow rates. More permeable parts of an aquifer with larger hydraulic conductivities may in concept be more vulnerable to more rapid flow of potential contaminants toward a sampled or supply well than are less permeable parts of that aquifer. Aquifer-test results cited from the GM-BVA indicate a range of hydraulic conductivities from about 0.33 to 2,500 feet per day (ft/d; Dumouchelle and others, 1993; Dumouchelle, 1998). Additional hydraulic characteristics summarized for specific parts of the GM-BVA hydrogeologic framework included those for valley fill (horizontal hydraulic conductivity ranging from about 0.3 to 500 ft/d; the ratio of horizontal to vertical hydraulic conductivity of 10:1; and porosity ranging from 0.15 to 0.25), till (horizontal hydraulic conductivity ranging from about 0.07 to 10 ft/d and porosity from 0.15 to 0.25) and adjacent bedrock (hydraulic conductivity ranging from about 0.003 to 5 ft/d) (Sheets, 2007). Well yields in the GM-BVA commonly exceed 1,000 gallons per minute, and larger yields are generated near the largest streams (Norris and Spieker, 1966; Dumouchelle, 1998).

Isotopic derived groundwater-age estimates also indicate the vulnerability of the GM-BVA to recent contamination. Groundwater-age estimates based on prior application of tritium and helium-3 dating techniques indicate that sampled wells in the GM-BVA with shallower depths of 0 to about 50 ft from the water table to the top of the open or screened interval had groundwater ages since recharge in all but one case that ranged from a few months to about 13 years: one well had an age of 32 years (Rowe and others, 1999, fig. 23, p. 44). Recharge area wells in the GM-BVA with depths of 51 to 130 ft from the water table to the top of the open or screened interval had groundwater ages since recharge that ranged from about 6 to 26 years (Rowe and others, 1999, fig. 23, p. 44). Estimates of the composite age of recharge that contribute to groundwater produced from a well can indicate the time between input of a possible contaminant at the water table and when it arrives at a sampled well (Eberts and others, 2013).

Well Selection and Related Characteristics

Wells used to collect groundwater samples for PFAS analysis were previously sampled by the National Water-Quality Assessment (NAWQA) Program as part of the Great and Little Miami River Basins Study Unit (Rowe and others, 1997). The NAWQA Program later became a project within the NWQP. The wells used were from a major aquifer study-unit survey network (22 private wells) and an agricultural land-use survey network (1 monitoring well; SH–75). Construction and location characteristics of those wells are summarized in table 3. The 22 private wells were sampled in 1999 to assess concentrations of inorganic and organic constituents in the GM-BVA in southwestern Ohio and southeastern Indiana (Rowe and others, 2004). Well SH–75 is a 2-inch polyvinyl chloride (PVC) cased monitoring well that was sampled to replace a study-unit survey well that could not be accessed in 2019 or 2020. Well SH–75 was most recently sampled in July 2000 (USGS, 2022). Well SH–75 was not sampled for PFAS analysis by method 2 because no NWQP sampling of that well was planned in 2019. Two of the private wells, GR–653 and GR–651, were sampled as substitutes for other wells that could not be sampled as planned because of a well closure and well pump incompatibility with PFAS sampling.

Wells sampled for this study had total depths below land surface that ranged from 21 ft at CL–275 to 101 ft at MT–1250 (table 3). Wells sampled were constructed of steel, galvanized iron, or PVC casing materials. Well screen materials and well completion details were documented for all 23 wells sampled (table 3). Eighteen wells completed with well screens had lengths that ranged from 2 to 11 ft and were constructed with steel, stainless steel, galvanized iron, or PVC (table 3). Five wells sampled had open hole completions, meaning the base of casing was left open against aquifer material (table 3). At BU–1101, the open hole was completed using gravel in the lowermost 6 ft of well casing that permitted groundwater to flow into the well and prevented intrusion of aquifer sediments.

Sampling Preparation and Collection Procedures

Preparation of sampling materials and collection methods for several types of groundwater and quality-control samples analyzed for PFAS and for inorganic constituents to characterize redox categories of groundwater are described in this section. Two different proprietary adaptations of the EPA 537.1 method by different laboratories, defined as methods 1 and 2, were used to analyze samples for PFAS concentrations in groundwater from the GM-BVA wells.

Groundwater was withdrawn from all sampled wells, except SH–75, for purging, observation of field water-quality parameters, and sample collection by an existing submersible pump used by the private well owner for their water supply. Samples produced from the submersible pumps were collected from a threaded spigot. Spigots and plumbing were inspected to ensure water was produced before any treatment or pressure tanks. Groundwater was withdrawn from well SH–75 for purging, observing field water-quality properties, and collecting samples using a bottom-fill high-density polyethylene (HDPE) bailer because that well was a standpipe with no submersible pump. An HDPE bailer was used to ensure that sampling materials used to collect groundwater were compatible with PFAS (Michigan Department of Environmental Quality, 2018, p. 19).

Potential interfering conditions were noted during each groundwater sampling event (table 3). Fluorocarbon tape at the sampled spigot was noted before sampling because of its potential to be an extraneous source of PFAS to groundwater sampled from that source. One well, CL–277, had visible fluorocarbon tape on a pipe joint at the wellhead (table 3). Other potential interfering conditions noted near sampling points or the wellhead for other sampled wells were small amounts of pesticide or insect spray stored at MT–1250 and CH–101, household cleaners near W–53, and a septic tank and a pavement sealer spill in 2019 at CL–275. No other visible evidence of potential interfering conditions was observed at the other wells sampled.

Groundwater was purged from each well and its associated plumbing and field water-quality parameters (specific conductance, water temperature, pH, dissolved oxygen, and nephelometric turbidity) of the purge water were observed to stabilize before samples were collected (Wilde, 2008; Gibs and others, 2012). One well volume from the 22 wells with existing submersible pumps was purged before stabilization of field water-quality properties was evaluated because those wells were in regular use before sampling (USGS, 2006). Three well volumes were bailed from well SH–75 before stability of field water-quality parameters was evaluated. Well purging continued from the first stabilization check until water-quality parameters stabilized to within 3 percent for specific conductance, 0.2 degrees Celsius for temperature, 0.1 standard units for pH, 0.2 milligrams per liter (mg/L) for dissolved oxygen, and within 10 percent for turbidity (Wilde, 2008; Gibs and others, 2012). For water bailed from well SH–75, turbidity was deemed sufficiently small to meet purge and sampling criteria when measured to be less than 10 nephelometric turbidity ratio units (NTRU), a standard for PFAS sampling as described in “Section 11” of Interstate Technology and Regulatory Council (2022) and consistent with USGS practice (USGS, 2006, p. 91).

Filtered groundwater samples were first collected from the coordinated NWQP sampling of 22 of the 23 wells. Analytes of interest from that sampling included several inorganic constituents, including nitrate plus nitrite as nitrogen (N), nitrite as N, ammonia as N, sulfate, manganese, and iron, plus tritium. These inorganic constituents were used to characterize the redox category of water at the time of sampling, and the isotope tritium was used to qualitatively classify groundwater age. Samples for inorganic and tritium analysis were collected, and inorganic samples were filtered and stored in accordance with USGS procedures before collecting samples for PFAS analysis (USGS, variously dated; McMahon and others, 2022b).

Groundwater and sequential replicate samples were collected from the private wells for PFAS analysis through new, precleaned fittings dedicated to each well and a stainless-steel fitting that was cleaned before sampling and between sampling each well. Fittings consisted of—

The threaded nylon adapter and HDPE tubing were disposed after each use. New ferrule sets for the compression fitting were installed and cleaned for each well. Fittings were cleaned by pouring solutions through the stainless-steel compression fitting, the new threaded nylon adapter, and a segment of HDPE tubing in the following sequence—

New disposable polyethylene gloves were worn during all cleaning steps except during the methanol rinse when new disposable nitrile gloves were worn. The cleaned equipment was double bagged after cleaning in new, standard use plastic bags and taken to each sampled well for use.

Samples for PFAS analysis were collected after other samples by disconnecting all equipment used for sampling inorganic constituents and replacing that equipment in the following sequence—

The sample processing chamber was used during sample collection and sample processing to prevent potential airborne contaminant sources (such as PFAS) from contacting the sample (Wilde and others, 2014). Upon completing sample collection, the flow of water was stopped. Fittings and tubing were disconnected from the well spigot, and the stainless-steel compression fitting was disassembled from the threaded nylon adapter. The adapter, HDPE tubing, and stainless-steel ferrules on the tubing were disposed; however, the stainless-steel connector used to join the HDPE tubing to the stainless-steel compression fitting was retained, cleaned, and reused for the next sample. This process was repeated for each well sampled.

Groundwater and Quality-Control Sample Types for Per- and Polyfluoroalkyl Substances Analysis

The different types of groundwater samples collected, quality-control samples prepared for PFAS analysis, and abbreviations used for each sample type in this report are described in this section. Two groundwater samples were collected for PFAS analysis at all wells except SH–75 using the same sampling procedures but were analyzed by different laboratories using 1 of 2 different adaptations of EPA method 537.1 (Shoemaker and Tettenhorst, 2018). Each sample type has a “-method 1” or “-method 2” suffix that indicates which method 1 or 2 adaptation of EPA method 537.1 was used to identify and determine concentrations of PFAS in the water sample (table 4). Laboratories and methods used for PFAS analyses are described in a later section titled “Laboratory Analysis Procedures.”

The coordinated sampling and analysis by methods 1 and 2 was intended to provide results from paired irreplicate samples so that a comparison of the differences in concentrations and number of PFAS detections in groundwater yielded from the two different methods could be made. The term “irreplicate” was used in this report to describe how paired samples collected sequentially from a well were used to understand the comparability of data yielded from analyses by different analytical methods applied by different laboratories, as distinct from replicate samples that were analyzed by the same analytical method and laboratory (Mueller and others, 2015, p. 7). Comparisons with several types of quality-control samples were used to verify results. Methods 1 and 2 were anticipated to have differing RLs and DLs for specific PFAS; details of those terms and differences are discussed with the analytical results. This use of paired irreplicate groundwater samples to assess the comparability of analyses by different laboratories or analytical methods was consistent with the definition of irreplicate samples and their use in USGS quality-control procedures (Mueller and others, 2015, p. 7). Details of laboratories and methods used for PFAS analyses are described in a later section titled “Laboratory Analysis Procedures.”

Groundwater-method 1 (GW-method 1) and groundwater-method 2 (GW-method 2) refer to groundwater-collecting methods used to represent groundwater quality at the time of sampling and whose results were analyzed for PFAS concentrations through different contract laboratories, each using a separate proprietary adaptation of EPA method 537.1 (method 1 or method 2) with different RLs and DLs for specific PFAS. The GW-method 1 samples were collected for this study, and the GW-method 2 samples were collected for the companion NWQP sampling (McMahon and others, 2022b).

Laboratory Analysis Procedures

Method 1 or 2 analysis of PFAS, respectively, were completed by different commercial laboratories. Methods 1 and 2 were separately adapted from EPA method 537.1 (Shoemaker and Tettenhorst, 2018) and used to quantify PFAS concentrations in non-drinking water samples. Analyses of PFAS by method 1 were performed by Bureau Veritas Laboratories, Mississauga, Ontario, Canada (formerly MAXXAM) using Bureau Veritas method CAM SOP–00894 (table 4), as adapted from EPA method 537.1 and with quality-control criteria referenced to appendix B, table B–15 and appendix C, table C–44 of the U.S. Department of Defense Quality Systems Manual version 5.2, as consistent with that in version 5.3 (U.S. Department of Defense, 2019). Analyses of PFAS by method 2 were performed with a different proprietary method by SGS North America Inc. (SGS-Environment Health and Safety, Orlando, Florida) using SGS standard operating procedure MS 014.9 (table 4), as adapted from the EPA method 537.1 and with quality-control criteria referenced to appendix B, table B–15 of U.S. Department of Defense Quality Systems Manual version 5.1 (U.S. Department of Defense, 2017). Details of method 2 analysis and validation are in McMahon and others (2022b).

Sample extractions for PFAS analyses were completed according to several steps adapted from EPA method 537.1 (Shoemaker and Tettenhorst, 2018) that were common to methods 1 and 2. The full volume of water in a sample bottle was extracted for PFAS at the laboratories by methods 1 and 2; for method 1, the sample volume extracted was 125 mL, and for method 2, the sample volumes extracted ranged from 240 to 290 mL. Each water sample was fortified with isotopically labeled internal standard PFAS extracted on a solid phase-extraction (SPE) cartridge, and the SPE cartridge was subsequently dried under vacuum. Internal and surrogate standard PFAS added to each sample included carbon-13, deuterium (hydrogen-2) or oxygen-18 isotopically labeled versions of most PFAS in table 1, with “n” being the number of atoms of that isotope in the compound. Sample bottle rinsates were also extracted on a SPE cartridge to yield surface adhering PFAS. The water sample and bottle rinse extracts were combined, carbon cleaned, brought to a standard volume with reagent and concentrated to a reduced standard volume to analyze a final “sample extract.”

Methods 1 and 2 used different adaptations of permissible liquid chromatography with quadrupole mass spectrometry columns, media, equipment, and operating conditions that were appropriate for each laboratory’s systems and were validated using methods sufficient to meet quality-control requirements of EPA method 537.1 (Shoemaker and Tettenhorst, 2018) to determine and quantify PFAS listed in table 1. Individual PFAS were analyzed by liquid chromatography with quadrupole mass spectrometry in methods 1 and 2 by injecting a standard volume of sample extract onto a liquid chromatography column where the PFAS were separated before detection by tandem mass spectrometry. Individual PFAS were identified by comparing mass spectra and retention times for compounds eluting from the column to those from internal standard PFAS used for calibration by each method. Each PFAS was quantified using an isotope-dilution process that involved adding an isotopically labeled internal standard or an appropriate internal standard to a sample, measuring the ensuing isotopic composition, and using those data to calculate the amount of analyte present in the sample. Percentage recoveries of the internal and surrogate standards from the sample extracts were used to evaluate the ability of each method to quantify PFAS in groundwater and quality-control samples.

Reporting limits for each PFAS determined by method 1 or 2 were defined as the smallest true concentration that each analytical method could reliably detect and recover from a sample at percentages ranging from 50 to 150 percent and at a 99-percent level of confidence (Shoemaker and Tettenhorst, 2018, p. 537.1–2). Detection limits for each PFAS determined by methods 1 and 2 were defined as the statistically calculated minimum concentrations that each analytical method could measure at a 99-percent confidence level and the value is greater than zero (Shoemaker and Tettenhorst, 2018, p. 537.1–3). The RLs and DLs can vary over time and were dependent on the compound being analyzed, water sample characteristics, laboratory preparation of the extract to be analyzed, and instrument characteristics (Winslow and others, 2006; Shoemaker and Tettenhorst, 2018). A method with a RL and a DL that detects smaller concentrations of a particular PFAS represents a more sensitive analysis for that substance.

Three types of PFAS analytical results are described in this report: Those with concentrations greater than or equal to the RL, those with concentrations between the RL and the DL, and those not detected in a sample. Concentrations of PFAS quantified at or above the RL were listed in data tables and figures as numeric concentrations without data qualifiers. Concentrations of a positively identified PFAS that were less than the RL but greater than the DL were reported in tables in this report with an accompanying footnote to indicate that the concentration is approximate, an annotation such as the “J” code used by EPA (2018). Non-detections of PFAS were reported as a value less than the RL of the method used for analysis.

Laboratory methods used for all 2019 and 2020 analytical results summarized by this report are listed in table 4. Filtered groundwater samples collected from 22 of the 23 wells by the contemporaneous NWQP-led sampling and by a separate sampling of well SH–75 on August 17, 2020, for the redox constituents were analyzed for nitrate plus nitrite as N, nitrite as N, ammonia as N, sulfate, manganese, and iron at the USGS National Water Quality Laboratory in Lakewood, Colorado. Tritium isotopic analyses from the NWQP-led sampling were analyzed at the USGS Tritium Laboratory in Menlo Park, California, using electrolytic enrichment and liquid-scintillation counting (McMahon and others, 2022b; Thatcher and others, 1977). Prior tritium-helium-3 isotopic analyses reported in Hinkle and others (2010) were analyzed at Columbia University-Lamont‐Doherty Earth Observatory, Palisades, New York, using methods described in Ludin and others (1998). Results of these 2019 and 2020 groundwater sample analyses are also reported in the USGS National Water Information System, as referenced to the station identification numbers in table 3 (USGS, 2022).

Evaluation of Quality-Control Samples and Comparisons

Concentrations of PFAS in quality-control samples were compared with PFAS concentrations in their closest in time groundwater samples that were analyzed using the same method to assess if sample-collection related interferences were present. Quality-control sample results were categorized as follows according to Ohio Environmental Protection Agency criteria (Ohio Environmental Protection Agency, 2012).

Concentrations of PFAS in paired irreplicate and sequential replicate, groundwater samples were compared for consistency using a combination of qualitative and quantitative methods. Examples of paired sample comparisons included:

Differences between concentrations of an individual PFAS in a water sample and its paired sample were evaluated where practical using the relative percent difference (RPD) statistic. An RPD statistic was computed when concentrations of an individual PFAS were detected in both paired samples above the RL of the analytical method applied to each sample. The RPD is the absolute value of the difference of the concentrations of a single constituent in two samples, divided by the average of the two concentrations, expressed as a percentage (U.S. Army Corps of Engineers, 2008; Mueller and others, 2015). The statistic was computed as

Comparisons of results from paired samples were classified according to the following categories.

Land Uses and Potential Sources Proximate to Sampled Wells

The maximum number of PFAS detected in a groundwater sample from each well was compared with two geographic characteristics proximate to each well that may describe possible PFAS sources; dominant land uses and potential facility and industry points of interest that may have used PFAS as of 2012. The comparison was intended to understand the relation of possible PFAS sources overlaying the GM-BVA to concentrations of PFAS in groundwater. Groundwater samples with detections of one or more PFAS in a sample were compared with groundwater samples with no detections of PFAS in any sample.

Land uses as of 2012 within a 0.3-mile buffer around a sampled well were considered proximate land uses, as defined using classifications published for 22 of 23 wells by McMahon and others (2022b), as derived using data from Falcone (2015). Land uses as of 2012 that were represented by comparisons in this study included agriculture, natural, and urban classes. Land uses within a radial distance of 0.3 miles around a sampled well previously were demonstrated to correlate with nitrate and several types of man-made organic compounds, as summarized in McMahon and others (2022b).

Potential facility and industry points of interest were considered proximate to sampled wells if they were within a 2-mile buffer around a sampled well. Potential facility and industry points of interest proximate to 22 of 23 wells sampled by this study were defined using data sources and distances from McMahon and others (2022b). The potential facility and industry points of interest that may have used PFAS as of 2012 were facilities defined as fire stations, wastewater treatment, defense, landfill, and public use airports and industries defined as metal coating or metal machining, paper production, petroleum products, plastics (resin), electronics, chemicals, paints, and cleaning. The 2-mile radial distance used by this study to assess points of interest relative to a sampled well was similar to the distance used by other studies to compare land use and pesticide detections in groundwater (Kolpin and others, 1995; Worrall and Kolpin, 2004).

Prior published data were not available for land uses and facility and industry points of interest data proximate to well SH–75. Land uses within 0.3 mile of well SH–75 were qualitatively classified by this study into agriculture, natural, and urban classes using imagery referenced from Google Earth (2022). Each land use class for well SH–75 was 50 percent or more of the land use within a 0.3-mile radial distance of the well, less than 50 percent, or if not present, zero percent. Facility and industry points of interest within 2 miles radial distance from well SH–75 were also visually classified and counted by this study using imagery referenced from Google Earth (2022).

Groundwater-Age Estimates and Redox Categories

Groundwater-age estimates based on 1999–2000 data that described the dates of groundwater recharge produced from the sampled wells (Hinkle and others, 2010) and tritium-based groundwater-age categories from samples collected in 2019–2020 (McMahon and others, 2022b) were used to understand aspects of PFAS detection and non-detection in groundwater from the GM-BVA. Groundwater-recharge dates were used to evaluate whether the sampled wells produced water that substantially recharged before or after the introduction of PFAS into common use. PFAS detections in GW-method 1 and GW-method 2 samples were also evaluated to understand if they coincided with groundwater-recharge dates within the general period of common use for that compound. Tritium-based groundwater-age categories from 2019–2000 data for 22 of the 23 wells used for PFAS sampling by this study were compared with the groundwater-age estimates derived from the 1999–2000 data to identify major differences.

Concentrations of tritium (in 22 of 23 wells) and its radioactive decay daughter product helium-3 (in 17 of 23 wells) in prior (1999) groundwater samples had been used to estimate the age of groundwater produced from the GM-BVA (Hinkle and others, 2010). The groundwater-recharge date relative to samples collected by this study was computed for 22 sampled wells by subtracting the groundwater-age estimate from a 1999 or 2000 sample from the year that a sample was collected for PFAS analysis from the same well (2019 or 2020).

Recharge ages computed from the 1999 or 2000 data were also checked by comparing with a tritium-based groundwater-age category computed from tritium concentrations reported from the 2019–20 companion NWQP sampling (McMahon and others, 2022b). Tritium-based groundwater-age categories were classified as “pre-modern” recharge from before 1953, “modern” recharge from 1953 or later, or mixed-age recharge because they comprised pre-modern and modern components (Lindsey and others, 2019; McMahon and others, 2022b).

The tritium-helium-3 and tritium-based, piston-flow estimates of groundwater age cited in this report from the 1999 and 2000 sampling of these wells were described by Hinkle and others (2010) as simplified representations of the tracer data. Piston-flow estimates of groundwater ages since recharge are based on the potentially limiting assumptions that tracer transport is advective and that no mixing occurs in groundwater between its recharge below the water table and where it was withdrawn from each well during sampling (Hinkle and others, 2010). Additional uncertainties with tritium-helium-3 and tritium-based, piston-flow estimates of groundwater age can originate from the terrigenic (natural) sources of tracers, the spatially varied atmospheric tracer concentrations, and the incomplete understanding of recharge mechanisms and were addressed in analyses of similar data from a subset of the study area.

Prior analyses of tritium-helium-3-based groundwater-age estimates from a part of the GM-BVA in the Dayton area yielded chemically and hydrologically reasonable results, with groundwater-age estimates increasing with depth in the aquifer and along regional groundwater flow paths and decreasing with distance between a recharge area and pumping centers in the GM-BVA (Rowe and others, 1999; Shapiro and others, 1998). Overall close agreement was identified between an estimated tritium-input function for rainwater in southwestern Ohio and the sum of tritium and tritiogenic helium-3 in most groundwater samples (Rowe and others, 1999; Shapiro and others, 1998).

Groundwater-recharge dates for samples collected in 1999 from wells BU–1106 and CL–278 were estimated solely using their tritium concentrations because those samples contained terrigenic helium (BU–1106) or had imprecise helium determinations (CL–278) that prevented computation of a tritium-helium-3-based age estimate (Hinkle and others, 2010). Groundwater-recharge dates that were estimated using only tritium concentrations were classified as modern and post-1952 in age but may also have contained some fraction of pre-modern water, depending on the actual age of the modern water in the mixture and the date of sample collection (Lindsey and others, 2019).

Redox categories and processes were classified for groundwater samples collected from the 23 wells using a framework defined by McMahon and Chapelle (2008). Concentrations of dissolved oxygen, sulfate, several N species, manganese, and iron in groundwater samples reported in this study and from a prior 1999–2000 sampling of the same wells (USGS, 2022) were classified through a ranking procedure into redox categories of oxic, suboxic, anoxic, or mixed (Jurgens and others, 2009; McMahon and Chapelle, 2008). Comparisons of redox categories from samples of this study and from the prior 1999–2000 samples from the same wells were used to indicate differences in overall redox processes over time.

The redox categories were used to identify processes in groundwater that could potentially affect aerobic or anaerobic transformation of PFAS or their precursor compounds. Although PFAS are thought to persist under many environmental conditions, groundwater with aerobic or oxic redox categories can be associated with processes that facilitate transformations of PFAS precursor compounds, such as those in aqueous firefighting foams, into terminal degradation products such as PFOS and PFOA in groundwater (Weber and others, 2017). A perfluorooctane sulfonamide in commercial fabric treatment products, N-ethyl perfluorooctanesulfonamidoethanol, was biotransformed by contact with aerobic wastewater treatment sludge in microcosm experiments into N-ethyl perfluorooctanesulfonamidoacetate (EtFOSAA; Boulanger and others, 2005). Compounds such as PFOS, PFBS, and 6:2 fluorotelomersulfonate (6:2 FTS), however, were resistant to aerobic and anaerobic microbial degradation in the months and years long microcosm tests conducted with wastewater sludge (Ochoa-Herrera and others, 2016). Complex anaerobic microbial processes associated with iron and N reducing redox processes have been observed to defluorinate PFOA and PFOS in laboratory microcosm experiments that used much larger concentrations of PFAS and other constituents than are typically in environmental samples (Huang and Jaffe, 2019).

Quality-Control Sample Results

No PFAS were detected in analyses of pre-sampling blank (PSB) and field-blank (FB) quality-control samples (tables 8–10), indicating that source-solution water used to rinse equipment and pre-sampling equipment preparation protocols did not contribute to PFAS results in GW-method 1 and GW-method 2 samples (tables 8–10). No PFAS were detected in pre-sampling equipment blanks (PSB–E) prepared after equipment cleaning or in pre-sampling source-solution blanks (PSB–S) that were prepared and submitted for methods 1 and 2 analyses before any groundwater sampling was done (tables 8–10). Similarly, no PFAS were detected in FB samples prepared after equipment cleaning or before sampling at six wells, including six FB-method 1 and 2 FB-method 2 samples. No PFAS were detected in a source-solution FB submitted for method 1 analysis (FB–S-method 1). The lack of PFAS detections in analyses of PSB–S-method 1, PSB–S-method 2, and FB–S-method 1 samples indicated that PFAS-free and organic-free source-solution water used to rinse equipment and other equipment cleaning protocols did not contribute to results measured in GW-method 1, GW-method 2, and paired replicate samples.

Field blank samples prepared identically to a GW-method 1 sample from well CL–275 on April 20, 2020, and identically to a GW-method 2 sample from well MT–1255 on August 5, 2019, each had no detections of PFAS (tables 8–10). Those wells had produced groundwater samples in 2019 with concentrations of 3 to 4 different PFAS (CL–275, GW-methods 1 and 2 samples) and two different PFAS (MT–1255, GW-method 2 sample, tables 5–7), respectively. The field blank result from well MT–1255 indicated that cleaning sampling equipment after groundwater sampling was effective at preventing carry over of detectable PFAS residues between samples.

Detections or non-detections of PFAS in groundwater were also compared with other geographic, hydrogeologic, and water-chemistry-based characteristics that may affect PFAS concentrations. Characteristics compared included land use and possible PFAS sources at specified reference distances from the wells; the age of groundwater recharge to sampled wells as compared with general periods of PFAS use; redox characteristics of groundwater; and well, water-level, and field water-quality characteristics of the sampled wells.

Comparison of Reporting and Detection Limits for PFAS Analyzed by Methods 1 and 2

Method 2 had smaller RLs for 22 of 24 PFAS and smaller DLs for all 24 PFAS compared with method 1 (table 11). The smaller DLs made method 2 an overall more sensitive method to detect PFAS in groundwater and quality-control samples. Exceptions to that generalization included RLs for analyses of N-methyl perfluorooctanesulfonamidoacetate (MeFOSAA) and EtFOSAA. Analyses of MeFOSAA by method 1 in water samples from 2019 had slightly smaller but similar RLs as compared with those of method 2 (table 11). The RLs for analyses of EtFOSAA in water samples by methods 1 and 2 from 2019 were equivalent in 18 groundwater samples and were smaller for method 1 than those of method 2 in 4 groundwater samples (tables 5–7 and 11). Data used to prepare table 11 are reported in Buszka and others (2023).

The smaller RLs for all PFAS and the smaller DLs for most PFAS analyzed by method 2 likely relate, in part, to the larger sample volumes extracted by method 2 that ranged from 240 to 290 mL and were from 1.92 to 2.32 times greater than method 1 (125 mL). The larger sample volume extracted by method 2 would be more likely to yield a larger and potentially more detectable and quantifiable mass of a PFAS than from the smaller sample volume extracted by method 1. This difference in extraction yield would account for part of the difference between the RLs and DLs of PFAS targeted for analysis between methods 1 and 2. Other method differences may account for the remaining differences and greater sensitivity of method 2 relative to method 1 for most PFAS. Similarities of RLs for method 1 and 2 analyses of MeFOSAA and EtFOSAA and of DLs for the same may indicate other analytical factors that affect recovery of those two PFAS from water samples.

The RLs for method 2 varied among all analyses by values ranging from 3.4 ng/L for 18 different PFAS to 21 ng/L for MeFOSAA and EtFOSAA (table 11). The DLs for method 2 also varied among all analyses by values ranging from 0.86 ng/L for 15 different PFAS to 3.4 ng/L for MeFOSAA and EtFOSAA (table 11). In comparison, RLs and DLs for method 1 analyses were consistent for samples analyzed in 2019 (table 11). Common reasons for variation in RL and DL are because of slight differences in several factors, including sample matrix composition, extraction efficiency, concentrations in laboratory fortified blank samples used to quantify compound recoveries, and instrument operating conditions (Shoemaker and Tettenhorst, 2018).

Results of PFAS Analyses of Groundwater by Methods 1 and 2

Two groundwater samples from the GM-BVA had concentrations of PFOS and PFOA in one sample each that exceeded their EPA IHA guidance (tables 2 and 12). A PFOS concentration of 1.9 ng/L in a GW-method 2 sample from well CL–275 and a PFOA concentration of 2.1 ng/L in a GW-method 2 sample from well BU–1106 were considerably greater as a percentage than their EPA IHA guidance by about 9,500 and 52,500 percent, respectively (table 12). In comparison, concentrations of PFOS in the GW-method 2 sample from well CL–275 and of PFOA in the GW-method 2 sample from well BU–1106 were less than their respective Ohio action levels as of 2019, of about 2.7 and 3.0 percent, respectively (table 12).

Detection of PFOS or PFOA in groundwater at or less than concentrations defined by the EPA IHA guidance was not possible using analytical methods 1 or 2. For context, the EPA IHA guidance as of June 2022 for PFOS (0.02 ng/L) and PFOA (0.004 ng/L) were also 65 and 215 times less, respectively, than the smallest DLs for PFOS (1.3 ng/L) and PFOA (0.86 ng/L) reported for method 2—the more sensitive of the two methods used in this study. The EPA IHA guidances for PFOS and PFOA in drinking water identify EPA 537.1 as a quality assured analytical method developed by that agency to monitor for those PFAS in treated groundwater (EPA, 2022d, 2022e). When groundwater was sampled for this study in 2019 and 2020, the application of methods 1 and 2, as adaptations of EPA method 537.1, represented feasible approaches to analyze water samples for the presence and concentrations of targeted PFAS, including PFOS and PFOA. The lack of PFOS and PFOA detections in groundwater from 21 of the 23 wells sampled, however, did not rule out the potential presence of PFOS or PFOA at concentrations less than method 1 or 2 DLs that would also exceed the EPA IHA guidance.

Aside from the PFOS and PFOA detections described above, other PFAS were either not detected or were detected in concentrations less than Ohio action levels or Federal health-risk-based guidance (tables 5–10 and 12). Nine sampled wells had one or more GW-method 1 or GW-method 2 samples with concentrations of one or more PFAS that were compared with an Ohio action level or an EPA health advisory, including wells BU–1106, W–52, GR–650, GR–651, MT–1251, MT–1255, MT–1250, CL–275 and CH–103 (table 12). A concentration of 16 ng/L of PFHxS in the GW-method 2 sample from CL–275 was the largest PFAS concentration detected by this study. That PFHxS concentration was about 11.4 percent of its Ohio action level of 140 ng/L and was the largest percentage of any PFAS analyzed by this study relative to its drinking water guidance (table 12; Ohio Environmental Protection Agency and Ohio Department of Health, 2019). The most detected PFAS was PFBS, which had concentrations in groundwater samples from eight wells that ranged from 1.0 to 8.0 ng/L (table 12). Those concentrations ranged from 0.05 to 0.40 percent of their of their EPA health advisory of 2,000 ng/L (tables 2 and 12). Four additional PFAS with concentrations in a groundwater or a replicate groundwater sample and that did not have an Ohio action level or Federal health-risk-based guidance included perfluoropentanesulfonate (PFPeS) from well CL–275, perfluorobutanoate (PFBA) from well GR–650, perfluoropentanoate (PFPeA) from well CH–100, and PFOSA from well MI–203 (table 12). Water samples from two wells had detections of at least one PFAS at a concentration greater than their method 2 RL, but no PFAS was detected at a concentration greater than their method 1 RL (tables 5–7; fig. 4).

No detectable PFAS concentrations were identified in an analysis of a follow-up April 21, 2020, GW-method 1 sample from CL–275 (tables 5–7). For reference, the July 9, 2019, sample from CL–275 had the largest number of PFAS detected in groundwater samples (PFBS, PFPeS, PFHxS, and PFOS) from this study. Non-detection of these PFAS in follow-up GW-method 1 and replicate (Rep–GW-method 1) samples from CL–275 on April 21, 2020, indicates that the 2019 results represented a transient detection in groundwater that did not persist with time (tables 5–10). A possible explanation for the differences in PFAS concentrations between the 2019 and 2020 samples from CL–275 relates to groundwater level differences between the two samples and is evaluated later in this report in the section titled “Comparison with Groundwater Levels, Well Characteristics, and Field Water-Quality Determinations.”

Eight PFAS targeted for analysis were detected in GM-BVA groundwater or in a paired replicate sample that were analyzed by methods 1 and 2 (table 12). Eleven of the twenty-three wells sampled in 2019 had from 1 to 4 PFAS detected in one or more groundwater samples or in a paired replicate sample and analyzed with methods 1 or 2 (fig. 5). The PFAS detected in these samples included PFBS in 8 wells and 9 samples, PFHxS in 4 wells and 5 samples, PFPeS in 1 well and 2 samples, and PFOS, PFBA, PFPeA, PFOA and PFOSA in 1 well and 1 sample each (table 12; fig. 5). Eight wells had from 1 to 3 PFAS detected only in samples analyzed by method 2, two wells (CH–100 and MI–203) had one PFAS detected only in samples analyzed by method 1, and one well (CL–275) had multiple PFAS detected in samples analyzed by both methods 1 (three PFAS detected) and 2 (four PFAS detected) (table 12; fig. 6).

Results from well CL–275 GW-method 1 and GW-method 2 analyses demonstrated the capability of both methods to yield similar concentrations of PFAS compounds when the concentrations were greater than the DLs of methods 1 and 2 (table 12). Similar PFAS concentrations were yielded from the July 9, 2019, GW-method 1 sample from well CL–275 (PFBS, 7.8 ng/L; PFPeS, 8.1 ng/L; and PFHxS, 14 ng/L) and from its paired irreplicate GW-method 2 sample (PFBS, 8.0 ng/L; PFPeS, 7.8 ng/L; and PFHxS, 16 ng/L) (table 12). Relative percent difference statistics computed for PFAS concentrations from the July 9, 2019, GW-method 1 and GW-method 2 paired irreplicate samples from well CL–275 were 2.5 percent for PFBS, 3.8 percent for PFPeS, and 13.3 percent for PFHxS. The similarity of concentrations yielded from analysis of paired irreplicate samples from well CL–275 by methods 1 and 2 demonstrated the capability of both methods to reproduce PFAS concentrations that were greater than their respective DLs.

More PFAS were detected in GW-method 2 samples than GW-method 1 samples because method 2 had smaller RLs and DLs for those compounds (table 12). The PFBS concentrations that were detected only in GW-method 2 samples included samples from seven wells (W–52, GR–650, GR–651, MT–1251, MT–1255, MT–1250 and CH–103; table 12). Other PFAS compounds with concentrations detected only in GW-method 2 samples and not in paired GW-method 1 samples were PFHxS from three wells (GR–650, MT–1255, and MT–1250); PFOS from well CL–275 (2019 sample only); PFBA from well GR–650; and PFOA from well BU–1106 (table 12). As explained earlier, the larger sample volumes extracted by method 2 would be expected to yield larger masses of PFAS and enhance their detection relative to the smaller sample volumes extracted by method 1. The PFAS compounds that were only detected in their GW-method 2 samples each had concentrations that were less than their DLs in method 1. The difference in PFBS concentrations between the GW-method 2 and GW-method 1 samples from GR–651 may also relate to the 19-day difference between their collection dates.

Sixteen of twenty-four PFAS targeted for analysis were not detected in GM-BVA groundwater samples analyzed by either method 1 or 2 (tables 5–10). The PFAS not detected in any groundwater or paired samples by methods 1 or 2 were all 3 fluorotelomer compounds (4:2 fluorotelomersulfonate [4:2 FTS], 6:2 FTS, and 8:2 FTS); 2 sulfonamide compounds (MeFOSAA and EtFOSAA); 3 sulfonate compounds (perfluorodecanesulfonate, perfluoroheptanesulfonate, and perfluorononanesulfonate); and 8 carboxylate compounds (perfluorohexanoate, perfluoroheptanoate, PFNA, perfluorodecanoate, perfluoroundecanoate, perfluorododecanoate, perfluorotridecanoate, and perfluorotetradecanoate) (tables 5–10). Non-detections of PFAS were also confirmed by results from GW-method 1 and GW-method 2 paired irreplicates from 22 wells or from comparison of GW-method 1 or GW-method 2 results with their paired Rep–GW-method 1 samples at 5 wells and Rep–GW-method 2 samples at 2 wells, respectively (tables 5–10).

No PFAS were detected in GW-method 1 or GW-method 2 samples from well CL–277 (tables 5–7). Well CL–277 was identified during inspection as having potentially interfering fluorocarbon tape on a pipe joint leading to the sampled spigot (table 3).

Two wells (CH-100 and MI-203) had PFAS concentrations yielded from analysis of their GW-method 1 samples that were greater than their detection limit but were classified as qualified results because they differed from results in paired Rep–GW-method 1 samples or from paired irreplicate GW-method 2 samples (table 12; fig. 6). A concentration of 5.4 ng/L of PFPeA in the GW-method 1 sample from well CH–100 was considered to be a qualified result because there was no corresponding detection of PFPeA in the paired Rep–GW-method 1, GW-method 2, and Rep–GW-method 2 samples from well CH–100 collected on the same date, despite the Rep–GW-method 1 analysis having the same DL and the GW-method 2 and Rep–GW-method 2 analyses having smaller DLs (table 12). Similarly, a PFOSA concentration of 8.7 ng/L in a Rep–GW-method 1 sample from well MI–203 was classified as a qualified result because its paired GW-method 1 sample from the same well did not have a similar PFOSA detection or concentration, despite having the same DLs (table 12).

Potential causes of differences in paired sample results from wells CH-100 and MI-203 could include short-term variability in groundwater quality, sample handling, laboratory recovery differences, or laboratory analytical conditions. Internal standard recoveries for methods 1 and 2 of the carbon-13 PFPeA surrogate were within acceptable limits for all samples collected from well CH–100, including Rep–GW-method 1 (78 percent), GW-method 1 (84 percent), GW-method 2 (93 percent) and Rep–GW-method 2 (98 percent). Carbon-13 PFOSA surrogate recoveries may explain the PFOSA detection differences between the Rep–GW-method 1 sample and paired GW-method 1 sample as recoveries of carbon-13 PFOSA surrogate were greater for the Rep–GW-method 1 sample (106 percent) in which PFOSA was detected than for the GW-method 1 sample (79 percent) in which PFOSA was not detected. Surrogate recoveries are available in a data release by Buszka and others (2023).

Concentrations of PFAS in GM-BVA groundwater as determined from GW-method 2 samples were considered verified results based on two comparisons with paired samples. Results of analyses from GW-method 1 and Rep–GW-method 1 samples collected from well CL–290 on August 13, 2019, and in follow-up GW-method 2 and Rep–GW-method 2 samples from the same well on March 4, 2020, all detected no PFAS (tables 5–10). Results of PFAS analyses for GW-method 2 and Rep–GW-method 2 samples collected from well CH–100 on August 6, 2019, also had no PFAS detections (tables 5–10).

The wells in the GM-BVA with PFAS concentrations in GW-method 2 samples that were less than a method 2 RL mostly did not coincide with sites that had been planned for Rep–GW-method 2 samples to be collected (tables 8–10 and 12). Analyses of the GW-method 1 samples from those same wells did not detect the same PFAS as detected by the GW-method 2 samples, likely because of the lack of sensitivity of method 1 at those smaller concentrations (table 12).

Interpretation of PFAS concentrations in GM-BVA groundwater depended on evaluations of the integrity of wells used to collect the water samples. Well integrity was assessed by a prior review of driller records as reported to the State of Ohio (Ohio Department of Natural Resources, 2022), sampling wells that reflected more recent construction and documentation practices (in this case, 1982 or later, table 3), inspecting visible characteristics at each wellhead, and by thorough pre-sampling purging before sampling each well. Domestic wells sampled for this study were selected using criteria described by Lapham and others (1995, 1997) and had previously been installed by water-well driller representatives of the private landowner. Well SH–75 was not a domestic well but was an observation well drilled by the USGS in 2000 to conform to NWQP protocols for sampling. Pre-sampling evacuation of at least three well volumes of water and the assurance that field-measured water-quality properties in the produced water had stabilized before sampling helped to minimize short-term (days to months) human-affected well integrity issues and enhance the likelihood that each result represented ambient groundwater quality.

Detailed results of groundwater PFAS analyses are summarized in the following list.

Results from well CL–275 indicate that repeated sampling of a well on multiple dates and analysis of those samples using an analytical method with sensitive RLs and DLs, such as method 2, are needed to assess the persistence and fluctuations of PFAS concentrations relative to health guidance, changing sources, and hydrologic conditions. Results from this study indicate the benefits of verifying PFAS concentrations in groundwater, such as collection and comparison of results from paired irreplicate, sequential replicate, and other quality-control samples.

Comparison of PFAS Detections with Land Use and Potential Facilities of Interest Proximate to Sampled Wells

Of the 23 wells sampled, agricultural land use as of 2012 was predominant (greater than 50 percent) within 0.3 mile of 12 wells, and urban land use was predominant within 0.3 mile of 11 wells (table 13; fig. 7). Agricultural land-use percentages near wells where agriculture was the largest proximate land use ranged from 50 percent or more at SH–75 to 99.1 percent at CH–102 (table 13). Urban land-use percentages near wells where urban land was the largest proximate land use ranged from 46.8 percent at W–53 to 100 percent at GR–651 and MT–1255 (table 13).

The most common potential facility points of interest that may have used PFAS as of 2012 within 2 miles of the sampled wells were fire stations, identified as proximate to 12 wells, and wastewater treatment facilities, identified as proximate to 11 wells (table 13). The most identified industry points of interest that may have used PFAS as of 2012 within 2 miles of eight sampled wells were metal coating or machining plants (table 13). Wells that had the most facility and industry points of interest nearby were MT–1255 with 10 sites, CH–103 with 5 sites, and H–151, MT–1250, and CL–275 with 4 sites each (table 13).

Results indicate that urban land use proximate to a well is a factor to assess when selecting wells for PFAS sampling. Wells with PFAS detected in a groundwater sample were more likely to have urban land within 0.3 mile as the largest percentage of all land uses within 0.3 mile (fig. 7). Six of nine samples from wells with more than about two-thirds (66.7 percent) urban land within 0.3 mile had concentrations of 1 to 4 PFAS detected in one of their groundwater samples (GR–650, GR–651, MT–1251, MT–1255, MT–1250, and CL–275; fig. 7). In comparison, 5 of 15 wells with less than about two-thirds (66.7 percent) urban land within 0.3 mile had samples with one or more PFAS detection.

The 6 wells with concentrations of PFAS detected in samples and that had more than 66 percent of urban land use as of 2012 (GR–650, GR–651, MT–1251, MT–1255, MT–1250, and CL–275) also had from 3 to 10 facility or industry points of interest within 2 miles or less that may have used PFAS as of 2012 (table 13 and fig. 7). In contrast, wells H–151 and CL–290 had no PFAS detected in samples and had 4 and 3 facility or industry points of interest within 2 miles or less that may have used PFAS, respectively (tables 5–7 and 13). The PFAS concentrations in a sample from well CH–103 coincided with five facility and industry points of interest within 2 miles of the well in an area that otherwise had only 3.2 percent urban land use within 0.3 miles of the well (fig. 7). In contrast, wells MI–203 and CH–100, which had qualified detections of PFAS in one of their groundwater samples collected in 2019, had no facility or industry points of interest within 2 miles of those wells (fig. 7).

The utility of land-use classifications and facility and industry points of interest proximate to a sampled well as variables to compare with PFAS detections in GM-BVA groundwater is limited by the accuracy of those classifications as of 2012, changes in existing activities before and since 2012, and the establishment of new activities since 2012. The land-use data cited in classifications for sampled wells were from about 2012 (McMahon and others, 2022b) or were classified by this study for well SH–75 using imagery also from 2012 (Google Earth, 2022). The facility and industry points of interest data used by McMahon and others (2022b) to derive distances from sampled wells were from the EPA Enforcement and Compliance History Online and the U.S. Department of Homeland Security, Homeland Security Infrastructure Protection Gold 2012 databases. Those data may include facility and industry points of interest that no longer perform the function they were classified as doing as of 2012, as described in EPA (2022f).

Comparison of PFAS Detections with Groundwater-Age Estimates

Groundwater-age estimates of GM-BVA groundwater from prior tritium-helium-3-based results from 17 of the wells sampled by this study ranged from less than 1 to 28 years before sample collection (fig. 8 and table 14; Hinkle and others, 2010). Groundwater-age estimates for sampled wells are summarized in table 14. These results place the estimated ages of recharge contributing to GM-BVA groundwater produced by these wells during 2019 and 2020 sampling from about 1991 (28 years before sample collection) at well CH–103 to 2018 (less than 1 year) at well W–53 (table 14). Groundwater-age estimates also indicate that water from wells sampled by this study had infiltrated into and recharged the water table in the GM-BVA within the overall 1947–present (2022) period of common use of PFAS or the environmental presence of many PFAS (fig. 9). The common use or environmental presence of PFAS and how it pertains to comparison with GM-BVA groundwater data is based on a generalized timeline of development, production process, or common use of select PFAS compiled for this study, as adapted from figure 2.1 in Interstate Technology and Regulatory Council (2022) and other references summarized in figure 9. The groundwater-recharge date range of 1991 to 2018 for the 17 wells sampled with tritium and helium-3 also coincided with newer uses of several PFAS, including PFBS, PFHxS, and fluorotelomers (fig. 9).

Groundwater samples from 2019 with detectable PFBS and PFHxS were from wells with groundwater-recharge dates that indicated sources of those compounds were from about 1991 to 2016. Groundwater samples with PFBS concentrations from eight wells (W–52, GR–650, GR–651, MT–1251, MT–1255, MT–1250, CL–275 and CH–103) had groundwater-age estimates that ranged from 1991 to 2016 (table 14). Those ages were sufficiently modern to coincide with the possible environmental presence of PFBS as a PFAS byproduct (wells CH–103 and GR–650) or its post-2002 use as an alternative to PFOS and other PFAS (wells W–52, MT–1255, CL–275, GR–651, MT–1251, and MT–1250; fig. 9). Before about 2002, PFBS was recognized as an impurity by-product of perfluorooctane sulfonyl fluoride production, with perfluorooctane sulfonyl fluoride being used to produce PFOS for commercial products (fig. 9; Bogdan, 2019). Products treated with PFOS and its derivatives, such as EtFOSAA, therefore, included PFBS and were considered a potential source of PFBS to the environment (Bogdan, 2019). Since about 2002, PFBS and its derivatives came into wider commercial use as a replacement for PFOS in flame retardants, fabric protectants, metal plating, surfactants, and some pesticide formulations (fig. 9; Bogdan, 2019). Wells GR–650 and CL–275 that had concentrations of PFHxS in 2019 groundwater samples also had post-2000 groundwater-recharge dates that coincided with the period of use of PFHxS as an alternative to PFOS (Minnesota Pollution Control Agency, 2021; Buck and others, 2011; fig. 9).

Non-detections of other individual PFAS in groundwater from wells sampled in this study are similarly meaningful to interpretations of PFAS detections. Wells with no detections of PFAS in groundwater samples but that had groundwater-recharge dates within the period of PFAS common use included H–151, BU–1101, W–53, GR–653, CL–278, CL–281, CL–290, CL–279, CL–277, CH–101, CH–102, and SH–75 (fig. 9). Wells with no detections of individual PFAS in groundwater samples but having modern groundwater-recharge dates concurrent with the period of PFAS common use indicate that they probably were not affected by a source of those PFAS (table 12).

Five of the twenty-three wells sampled by this study had a less precise groundwater-age estimate of less than 46 or 47 years before sample collection that were determined solely on tritium concentrations (Hinkle and others, 2010; fig. 6 and table 14). Of these wells, BU–1106 had a single detection of PFOA in a GW-method 1 sample (table 12). The estimated recharge date of 1973 for water from BU–1106 was within the period of PFOA common use (fig. 9). The remaining four wells in this category (CL–278, CL–279, CH–102, and SH–75) had no PFAS detections in their groundwater samples (fig. 9).

A groundwater-age estimate from well CL–275 was used to understand the likelihood of potential sources of PFBS, PFHxS, and PFPeS concentrations in GW-method 1 and GW-method 2 samples collected on July 9, 2019 (tables 12 and 14). Well CL–275 is in an area with principally urban land use within 0.3 mile of the well and four facility and industry points of interest within 2 miles of the well, including a wastewater treatment facility, a public use airport, electronics, and metal coating or machining (table 13). The well is also about 200 ft from the site of a tanker-truck accident that occurred about 12 days earlier on June 27, 2019 (WHIO–TV7, 2019). The truck was described as having leaked oil and possibly tar, such that the accident site was intended to be the subject of a post-incident cleanup of the leaked fluids. Well CL–275 has a very shallow total depth (21 ft, table 3) that indicates possible vulnerability to contamination from a very recent source. Well CL–275 had an estimated groundwater-recharge date of 2005 (table 14) that indicates that the June 27, 2019, accident was unlikely to be a source of contaminants in the July 9, 2019, groundwater samples. The 2005 groundwater-recharge date is within the period of common use for PFBS, PFHxS, and PFPeS and close to that of PFOS, which was also detected in the GW-method 2 sample (fig. 9). The data described in this paragraph indicate a greater potential for a pre-2019 source or sources to have contributed to PFAS detected in the July 9, 2019, groundwater samples from CL–275.

This interpretation of groundwater-recharge date in the 2019 sample from CL–275 depends on assumptions that groundwater produced by the well is affected only by advective flow from its source of recharge at the water table to its production at the sampled well and that very young, spill-related recharge did not otherwise mix with water in groundwater flow before it was produced from the well during sampling. The potential for older and younger flow paths to contribute to the overall age date in such a shallow well make it possible in concept for recharge with more recently introduced traces of contaminants to reach groundwater at the well screen along with groundwater from older flow paths. This type of interpretation of groundwater-recharge date and PFAS common use is therefore a likely explanation of PFAS concentrations in the 2019 sample from this well but may not rule out very young, spill-related sources of the compounds.

A potential limitation of using groundwater-age dates to interpret possible PFAS sources to a well is that the area contributing recharge to a well may not be static through time (Franke and others, 1998, p. 4). Local changes to the groundwater budget affecting these wells, such as changes in prevailing recharge to the aquifer during years of heavier precipitation, changes in aquifer stresses and flow rates from pumping at the sampled or adjacent wells, changes in local discharge, and changes in the well condition may affect sources, flow rates, and flow paths of groundwater to a well and its estimated age date at the time of sampling. The groundwater-age estimates used in this report to compare with PFAS detections in groundwater were from a 1999–2000 sampling of the same wells nearly 20 years before this study (Hinkle and others, 2010). Groundwater-age categories from McMahon and others (2022b) however corroborate the interpretation of data from Hinkle and others (2010) that groundwater sampled by this study in 2019 and 2020 was composed of modern, post-1952 recharge (table 14). The generally similar range of groundwater-age estimates from Hinkle and others (2010) and groundwater-age categories from McMahon and others (2022b) indicate that the estimated recharge dates of groundwater produced from these wells, and by inference their sources of recharge, were somewhat consistent through time.

The above results indicate that groundwater-age estimates from wells planned for PFAS sampling are a factor to assess when evaluating PFAS concentrations in groundwater. Resampling of age-dating constituents with a similar precision to the tritium-helium-3 method at the same time as PFAS sampling would also assist in understanding temporal changes in groundwater sources to sampled wells.

Comparison of PFAS Detections with Redox Categories and Processes in Groundwater

Groundwater samples in 2019 that had an oxic redox category were more likely to also have detections of one or more PFAS in a sample from that same well on the same date. Seven of nine wells had groundwater samples from 2019 with an oxic redox category and detections of one or more PFAS, including wells BU–1106, W–52, GR–651, MT–1250, MT–1255, CH–100, and CH–103 (fig. 10). In contrast, 4 of 11 wells had groundwater samples from 2019 that had an anoxic redox category and detections of one or more PFAS, including wells GR–650, MT–1251, CL–275, and MI–203 (fig. 10). Redox category and process classifications for sampled wells are summarized in table 15.

No apparent association between redox category and detections of PFBS and PFHxS in groundwater samples from 2019 was discernable. Five of eight wells that had a groundwater sample from 2019 with PFBS concentrations also had an oxic redox category, including wells MT–1250, W–52, GR–651, MT–1255, and CH–103 (tables 12 and 15). In contrast, 3 of 8 wells that had groundwater samples from 2019 with PFBS concentrations in one or both samples also had an anoxic redox category in groundwater sampled in 2019, including wells GR–650, MT–1251, and CL–275 (tables 12 and 15). Similar to the contrast just mentioned, wells MT–1250 and MT–1255 had PFHxS concentrations and an oxic redox category in groundwater sampled in 2019 and wells GR–650 and CL–275 had concentrations of PFHxS and an anoxic redox category in groundwater sampled in 2019 (tables 12 and 15). The redox categories for samples from well MT–1255 changed from anoxic in the 1999 sample to oxic in the 2019 sample, and the redox category for well MT–1250 was oxic in the 1999 and 2019 samples (table 15).

A redox category of oxic was indicated for groundwater sampled in 2019 and 2020 from 10 wells with dissolved-oxygen concentrations of 0.5 mg/L or greater (table 15; fig. 10). Redox processes affecting oxic redox category samples were characterized as oxygen reduction. Redox processes in anoxic category groundwater samples from 11 wells sampled in 2019 were mostly characterized as iron-reducing and sulfur oxidizing combinations, although well CL–275 produced samples with manganese reducing processes (table 15). Groundwater chemistry classified into an anoxic redox category had less than 0.5 mg/L of dissolved oxygen (Jurgens and others, 2009). A mixed-anoxic redox category consisting of nitrate reducing, iron-reducing, and sulfate reducing redox processes in the 1999 sample from CL–275 changed to an anoxic redox category in 2019 sample. A change to more reduced redox categories is consistent overall with the introduction of recharge containing a reducing agent, such as an organic-chemical contaminant or from changes in the sources of recharge to the well. A mixed redox category may arise when groundwater produced from the well originates from adjacent aquifer zones with differing chemical or microbial processes. A suboxic redox category was identified for a 2019 sample from well CH–101 (table 15 and fig. 10). A sample classified into a suboxic redox category has less than 0.5 mg/L of dissolved oxygen but lacks sufficient other data to classify a dominant redox process (Jurgens and others, 2009).

Changes in redox category were identified in the 1999 and 2019 groundwater samples from several other wells, including BU–1106 and MT–1255 (anoxic to oxic), MI–203 (mixed anoxic to anoxic), CL–279 (mixed-oxic-anoxic] to mixed-anoxic), and CH–101 (anoxic to suboxic) (table 15). Examples of processes that could affect changes in a redox category from anoxic to oxic or suboxic could include hydrologic conditions, such as an increase in very recent oxic groundwater recharge reaching the well screen, or changes in well integrity, permitting the entry of oxic groundwater to the well from shallower depth in the aquifer. Examples of processes that could change redox toward an anoxic category include decreased recharge to the aquifer surrounding the well with a deeper water table, decreased flow of oxic water to the well screen, development of microbial processes that consume oxygen in groundwater before it is produced from the well, or introduction of substances such as organic compounds whose biotransformation can consume oxygen as an electron acceptor.

Three PFAS that were possible terminal degradation products were detected in groundwater samples collected in 2019. These included PFOA in a GW-method 2 sample with an oxic redox category from BU–1106, PFBA in a GW-method 2 sample with an anoxic redox category from well GR–650, and a qualified concentration of PFPeA in a GW-method 1 sample with an oxic redox category from CH–100. Conditions characteristic of oxic and anoxic redox categories can indicate conditions favorable to biotic and abiotic degradation of some PFAS precursor compounds to form terminal degradation products such as PFBA, PFPeA, perfluorohexanoate, perfluoroheptanoate and PFOS (Interstate Technology and Regulatory Council, 2022; Weber and others, 2017). These possible terminal products can also be introduced into the environment from their direct use, so their presence in a sample is not solely diagnostic of transformation from precursor compounds.

Redox processes classified in water sampled in 2019 from wells BU–1101, CL–278, and possibly from well CL–279 indicated potential oxidation of ammonium to nitrite and reduction of ferric to ferrous iron. These wells produced groundwater with dissolved iron concentrations greater than 1,000 micrograms per liter (indicating a reduced redox category) and detectable concentrations of ammonia and nitrite (table 15). The 2019 groundwater sample from CL–275 had an anoxic redox category characterized by concentrations of nitrate, nitrite, ammonia, and manganese and had concentrations of several PFAS, including PFOS (tables 12 and 15). Although the above redox processes are superficially like the experimental conditions described for PFOA and PFOS defluorination by Huang and Jaffe (2019), overall concentration differences prevent evaluation of defluorination as an environmentally feasible process in GM-BVA groundwater.

Comparison with Groundwater Levels, Well Characteristics, and Field Water-Quality Determinations

Comparison of groundwater levels measured on the day of sampling in 2019 and PFAS concentrations in groundwater samples collected the same day yielded no apparent relation. Groundwater levels, characteristics of sampled wells, and field water-quality determinations are summarized in table 16. Well CL–275 had the shallowest groundwater level among wells sampled in 2019 at 1.83 ft below land surface and had PFAS concentrations in GW-method 1 and GW-method 2 samples (tables 12 and 16). Groundwater levels of the other 11 wells with PFAS concentrations in GW-method 1 or GW-method 2 samples ranged from 9.02 to 52.15 ft below land surface (table 16). In comparison, the groundwater levels measured on the day of sampling in 2019 in all 23 sampled wells ranged from 1.83 to 52.15 ft (table 16).