- Email: [email protected]
Calibration and validation of a novel passive sampling device for the time integrative monitoring of per- and polyfluoroalkyl substances (PFASs) and precursors in contaminated groundwater
Journal of Hazardous Materials 366 (2019) 423–431
Contents lists available at ScienceDirect
Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat
Calibration and validation of a novel passive sampling device for the time integrative monitoring of per- and polyfluoroalkyl substances (PFASs) and precursors in contaminated groundwater
T
Sarit L. Kaserzona, , Soumini Vijayasarathya, Jennifer Bräuniga, Linus Muellera, Darryl W. Hawkerb, Kevin V. Thomasa, Jochen F. Muellera ⁎
a b
Queensland Alliance for Environmental Health Sciences (QAEHS), The University of Queensland, 20 Cornwall street, Woolloongabba, Queensland, 4102, Australia Griffith School of Environment and Science, Griffith University, Nathan, Queensland, 4111, Australia
ARTICLE INFO
ABSTRACT
Keywords: Per-and polyfluoroalkyl substances (PFASs) Passive sampling Groundwater Microporous polyethylene Sampling rate
Per-and polyfluoroalkyl substances (PFASs) as key components in aqueous film forming foams (AFFF) have led to growing incidences of environmental contamination. The aim of this study was to investigate a novel diffusion based passive sampling device comprising of microporous polyethylene (PE) for the long-term time-integrative monitoring of PFASs in groundwater systems. PE passive samplers (PEs) were deployed for 83 d and calibrated at five AFFF impacted groundwater sites representing different PFASs concentration levels (ΣPFAS 0.001 to 0.1 ng mL−1). Grab samples were collected simultaneously. Linear accumulation of 12 PFASs (r2 ≥ 0.84) were observed in the PEs over 83 d and PFASs sampling rates were 2–5 mL d−1. Estimated mean half-times to equilibrium for PFASs ranged between 122 and 490 d. A separate validation study compared PEs and grab sampling during a 93 d field deployment, at seven groundwater sites near a fire fighting training ground. Seventeen PFASs were detected in PEs and fifteen in grab samples. PEs showed higher sensitivity for precursors (i.e. 4:2 FTS and FOSA). Time-weighted-average water concentrations across all validation sites for all PFASs determined from PEs were strongly correlated (r2 = 0.98) with grab samples, (within range 0.3–60 ng mL-1 PFOS). Results represent the first application of passive sampling technology for the quantitative assessment of PFASs in groundwater systems.
1. Introduction Per- and polyfluoroalkyl substances (PFASs) comprising compounds such as perfluoroalkylcarboxylates (PFCAs; CnF2n+1COO−) and sulfonates (PFSAs; CnF2n+1SO3−) have become ubiquitous and persistent environmental trace contaminants with over 3000 estimated to be used in various formulations on the global market [1,2]. Among the various industrial and domestic uses of these chemicals, it is estimated that approximately 200–300 occur in aqueous film-forming foams (AFFF) [1,3–5]. As a result of the physico-chemical properties such as aqueous solubility of some of the most widely distributed PFASs (i.e. PFOS, PFOA and PFHxS), aqueous phases play an important role in the environmental transport and fate of these chemicals [3,5–8]. A recent U.S study has identified that the main source of PFASs and highest risk factor for contamination in drinking water aquifers results from AFFF use (Department of Defense (DoD)) followed by PFASs manufacturing, waste streams i.e. landfills, AFFF use (airports) and fire-training areas
⁎
[8], emphasising the need for a diverse range of PFASs monitoring and management strategies. The collection of representative samples is a requirement for estimating PFASs concentrations over defined time scales and for accurate assessment of environmental burden [9]. In surface and wastewaters, the occurrence and extent of episodic events such as accidental spills or deliberate industrial release can be challenging to capture and adequately quantify. Contaminant levels in groundwater may show orders of magnitude difference between sampling sites, depending on factors such as hydraulic and geological conditions of the area, and the paths of chemical transport [3,8]. Leaching of PFASs into the environment from primary sources can persist for many decades [10,11]. However there is still a lack of understanding of what influences the mobility of PFASs in soils and groundwater [10]. Passive sampling techniques [12] have the potential to provide insight to groundwater contamination and hence become important complementary tools in PFASs pollution monitoring schemes. Some of the highlighted advantages of these techniques in this
Corresponding author. E-mail address: [email protected] (S.L. Kaserzon).
https://doi.org/10.1016/j.jhazmat.2018.12.010 Received 9 September 2018; Received in revised form 13 November 2018; Accepted 2 December 2018 Available online 07 December 2018 0304-3894/ © 2018 Elsevier B.V. All rights reserved.
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
regard include the ability to provide time-weighted average water concentration estimates over extended periods (i.e. days to weeks), relatively low limits of detection, cost effectiveness as well as ease and flexibility of deployment [13,14]. In 2012, the first calibrated passive sampler for the monitoring of a range of PFASs in surface waters, based on a modified Polar Organic Chemical Integrative Sampler (POCIS), was developed [15–17]. Results showed that all PFASs targeted were successfully sampled. However, equilibration of PFASs in the sampler was relatively fast with half-times to equilibrium (t1/2) for PFOS and PFOA ranging from 2.4 to 13 days. This meant samplers could only be operated in time-integrative mode for a relatively short period of time. Furthermore, the mass transfer and accumulation of PFASs in the samplers (i.e. sampling rates of PFASs) were shown to be influenced by deployment-specific conditions specifically, water turbulence [16], which could in turn cause the over- or under-estimation of water concentrations if laboratory-derived sampling rate data were employed instead of in-situ values [18]. Recently, a diffusion gradient in thin film (DGT) based passive sampler design (surface area 3.1 cm2) for organic chemicals (o-DGT) has been investigated as an alternative design for the sampling of polar organic chemicals [19]. Further development of the DGT sampler with XAD18 resin was employed for the detection of perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) in surface and waste waters. The reported diffusion coefficients (298 K) in the diffusive gel for this sampler were 4.37 × 10 − 6 and 5.08 × 10 − 6 cm2 s−1 (PFOA and PFOS, respectively). This, together with relatively high capacity for these compounds, suggested deployment times of several weeks were possible [20]. The use of a diffusion gel allows for a more accurate estimation of the mass transfer coefficient (i.e. sampling kinetics) of pollutants into the sampler with minimal effect from water turbulence or flow velocity on the sampling rates of chemicals. This is because the water boundary layer (WBL) thickness is typically much less than the gel thickness, resulting in a reduced influence of the WBL. However, challenges with this sampler design include relatively small sampling window surface areas and consequently lower sampling rates, susceptibility of the diffusion gels to depletion and biodegradation in some environmental systems which result in relatively short recommended deployment periods (2–4 weeks), and samplers that may be too big to fit into narrow groundwater bores [19,20]. Bopp et al. (2005) proposed a design suitable for groundwater deployment comprising of 8 cm long, 2 mm thick ceramic dosimeter tubes filled with sorbent (Amberlite IRA-743) for the passive sampling of VOCs [21]. In an aim to eliminate any WBL effects, a microporous diffusion layer replaced the membrane associated with POCIS. These tubes were trialled for monitoring flame retardants in river water [22] and for monitoring of cytostatic drugs in water [23]. However, it has been recognised that in some environmental matrices the ceramic pores of the sampler may be susceptible to erosion and/or clogging [23]. Furthermore the use of ceramic material for PFASs may be problematic, as PFASs demonstrate some affinity for silica (i.e. ceramics) negating its effectiveness as a suitable diffusion layer. Hence, sampling and laboratory material that are suitable for PFASs work often consist of polyethylene materials. In 2017, we successfully validated a similar tubular passive sampling device of the same dimensions manufactured from a cheap and robust microporous polyethylene (PE) enclosing a TiO2 receiving phase, for the monitoring of glyphosate and its transformation product aminomethylphosphonic acid (AMPA), both of which exist in ionized form in most surface waters [24]. The chemical diffusion characteristics and linear accumulation of such a sampler observed with these challenging anionic analytes raised the prospect of its successful application for PFASs, also environmental contaminants of contemporary concern. The overall aim of this study then was to develop a sensitive and robust monitoring tool for long term quantitative assessment of PFASs in contaminated groundwater. For this purpose, we investigate the microporous PE tube material in combination with an anion exchange
sorbent (Strata X-AW) as a passive sampling device. Firstly, the uptake kinetics and duration of the linear sampling mode were determined at five groundwater sites representing various PFASs contamination levels from historic AFFF activities. Any concentration dependency of the sampler and whether this may affect quantitative water concentration estimates was examined. We then apply the in-situ model calibration parameters obtained to estimate water concentrations of PFASs in groundwater sites located adjacent to a fire fighting training ground, and compare these to parallel grab sampling data. 2. Materials and methods 2.1. Microporous PE passive sampler calibration site description Microporous PE passive water samplers were deployed at five groundwater bores (denoted Sites 1–5) for periods of 14 or 15, 30 or 31, 40 or 41, 61 or 62 and 82 or 83 days. Site selection was based on (i) the ability to easily and safely approach and access sites (ii) a range of PFASs concentrations (i.e. low, medium and high) based on earlier studies. The contaminated sites sampled had restricted access and timepoints were chosen based on feasibility, relevance to users of the technology and access to sites. Site characteristics such as temperature, pH and electrical conductivity were collected at each sampling location throughout the sampling period (Table S1). 2.2. Microporous PE passive sampler validation site description The PE passive samplers were deployed at a total of 7 groundwater wells across 3 sites located adjacent to a fire training ground in Australia. Triplicate PE samplers were deployed for 93 days. Grab samples (500 mL) were collected at each site upon retrieval of the PE samplers from the wells. Information including depth of sampling, pH and salinity of water at sampling sites is included in Table S2. Grab samples (10 mL) from both validation and calibration sites were extracted as detailed by Bräunig et al. [3] (a brief description is provided in the Supporting Information (SI)). 2.3. Standards and reagents All equipment and materials (including glassware) were thoroughly rinsed with acetone (ACE) and methanol (MeOH) (Merck, Darmstadt, Germany; purity 99.8%) and allowed to dry prior to use. Ammonium acetate (> 97%, C2H7NO2) was obtained from ChemSupply (Gillman, SA, Australia). Ammonium hydroxide solution (NH4OH) was obtained from Sigma Aldrich. Water with resistivity > 18.2 MΩ cm (MQ) was obtained from a Millipore system. Samples were analysed for 29 PFASs and precursors (Detailed in the SI) including perfluoroalkyl sulfonates (PFSAs), perfluoroalkyl carboxylates (PFCAs), perfluorooctane sulfonamides (FOSAMs) and fluorotelomer sulfonates (FTSs). All analytical standards were purchased from Wellington Laboratories, Guelph, Canada. A native standard mix (PFAC-MXB, Wellington Laboratories) along with additional PFASs was used to prepare calibration standards in 40:60 MeOH:MQ (with 5 mM ammonium acetate) with a range from 0.1 to 100 ng mL−1. Mass labelled standards from Wellington Laboratories, MPFAC-MXA (13C4-PFBA, 13C2-PFHxA, 13C4-PFOA, 13C5PFNA, 13C2-PFDA, 13C2-PFUnDA, 13C2-PFDoDA, 18O2-PFHxS, 13C4PFOS) and M3PFBS (13C3-PFBS), M3PFPeA (13C3-PFPeA), M4PFHpA (13C4-PFHpA), M26:2 FTS (13C2-6:2 FTS) were used as surrogate internal standards. A mixture of M8PFOS and M8PFOA (13C8-PFOS and 13 C8-PFOA), were used as recovery standards. 2.4. Sample preparation and extraction 2.4.1. Passive sampler preparation Microporous polyethylene (PE) tubes (PALL, Germany) 0.8 cm in external diameter, 2 mm thick, with a pore size of 2.5 μm and 35% 424
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
concentration of the compound in water (ng L−1), RS the sampling rate (L day−1), t the time deployed (days) and mS the mass of the sorbent (kg). The half-life time to equilibrium (t1/2), describing the time it takes Cs to be half its equilibrium value is given by
porosity were cut into 4 cm lengths (Figure S1). Samplers were cleaned by first submerging them in a glass jar with acetone on a shaker for 24 h, followed by submergence in fresh acetone for another 24 h. This was followed by 2 x 24 h cleans in methanol. Samplers were left to dry in a fume hood. Each PE passive sampler was then packed with 400 ± 10 mg of sorbent phase material (Strata X-AW; Phenomenex, Australia). Both ends of the sampler were capped with 10 mm plastic round tubing insert caps (0.8–1.5 mm wall thickness; STOCKCAP, Australia). PE samplers were conditioned in methanol (1 x 24 h) on a shaker followed by 2 x 24 h in MQ water. All PE passive samplers were kept submerged in MQ water until deployment. Laboratory blank PE samplers were prepared alongside samplers (n = 6) and kept in the refrigerator until the processing and extraction of PE samplers.
t1/2 =
2.7. Quality control Laboratory blank passive samplers (n = 6) as well as laboratory and field blank grab samples (n = 7) were prepared, extracted and analysed in parallel with each batch of exposed samples. No PFASs were detected in blank PE samplers above LOQ while blank levels detected in grab samples are presented in Table S3. The method limit of detection was calculated as average blank level plus 3 times the standard deviation (SD). Instrument detection (LOD) and quantitation (LOQ) limits were calculated by multiplying the SD obtained from injecting the lowest calibration standard eight times by 3 and 10, respectively. Acceptable coefficients of variation (%CV) between triplicate PE passive samplers values were observed with < 28% for the majority of samples (Table S4). Grab samples showed %CV of ≤ 19% for intra-day grab samples, with the exception of 6:2FTS and 8:2FTS for which %CV was ≤ 60% due to trace levels amounts detected as reported in Table S3. Recovery of chemicals was verified by spiking blank and exposed samplers with PFASs surrogates prior to extraction, and recovery standards prior to analysis. Non-extracted side spikes (solvent blanks spiked with surrogates and recovery standards) were prepared in parallel to spiking and extracting exposed samples. These represent 100% recoveries and are essential in recovery correction calculations. Acceptable recoveries (typically between 42%–165%) were observed for PFASs passive samples (Table S4) with similar recoveries (30%–167%) for grab samples.
2.5. Analytical method All samples were analysed by high performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) using a Nexera HPLC (Shimadzu Corp., Kyoto, Japan) coupled to API5500 QTRAP mass spectrometer (Sciex, Melbourne, Australia) with an electrospray ionization (ESI) interface operating in negative ion mode. Chromatographic separation of the analytes was achieved with a Gemini C18 column (50 x 2.0 mm, 4 μm; Phenomenex, Torrance, CA), maintained at 45 °C, with a flow rate of 0.3 mL min−1 and injection volume of 5 μL. Mobile phases consisted of MeOH: water (1:99 v/v) (A), and MeOH: water (95:5 v/v) (B), with 5 mM ammonium acetate in both phases. An isolator column (Phenomenex) was included inline directly after the mobile phase mixing chamber to delay the elution of any solvent-derived background PFASs contamination. Data acquisition and processing was carried out using Analyst® TF 1.6 and MultiQuant™ software (Sciex).
3. Results and discussion 3.1. Concentrations of PFASs in groundwater samples Twelve PFASs were detected in grab samples across the five contaminated sites employed for calibration purposes, with eight consistently detected at all sites (Table S3). Mean PFASs groundwater concentrations (Cw) ranged from 0.01 ng mL−1 (for 6:2 FTS and 8:2 FTS) to 74 ng mL−1 for PFHxS. PFOS and PFHxS consistently exhibited the highest concentrations at all sites investigated followed by PFOA. ΣPFASs concentrations were 1.4 ng mL−1 (Site 1) < 3.7 ng mL−1 (Site 2) < 18.3 ng mL−1 (Site 3) < 26.8 ng mL−1 (Site 4) and < 157 ng mL−1 (Site 5) thus representing a range of contamination levels. 6:2 FTS and 8:2 FTS were detected only at sites with the highest ΣPFAS concentrations. Fluorotelomer sulfonates such as 6:2 FTS and 8:2 FTS have been reported to be degradation products of fluorotelomer-based AFFF foam (Ansulite), which replaced earlier AFFFs containing PFCAs/ PFSAs (3 M). A later transition to fluorine free foam (Solberg®) has been observed since 2010 [26]. No significant fluctuation of PFASs concentrations was observed between sampling days (%CV < 36% with the majority below 20% for all analytes and sites). Exceptions were found for 8:2 FTS at two sites (with %CV values of 49% and 60%) and PFHxS (with %CVs of 45% and 48%). Mean levels of these compounds represent concentration levels that were at the lower and upper range of the analytical calibration curve, respectively. Overall, concentrations of PFASs investigated were sufficiently stable to allow for the in-situ calibration of the PE passive sampling devices during the deployment period.
2.6. Data modelling Assuming a constant ambient aqueous phase concentration, the accumulation of ionizable and polar organic chemicals in the sorbent of polar passive samplers can be described using a one compartment, 1st order kinetic model (Equation 1). When uptake of analytes into the sampler is linear with time i.e. the sample is operating in kinetic mode, Equation 1 can be reduced to linear approximation model (Eq. (2)) [25].
Cs =
Cw Rs t ms
exp
Rs t ms Ksw
(3)
Values of Rs were estimated from Eq. (2), (by unweighted linear regression using GraphPad Prism 7.03). Ksw values were taken from Kaserzon et al. [15].
2.4.2. Passive sampler extraction Passive PE samplers retrieved from deployment were transferred into 15 mL Falcon tubes. The surrogate PFASs internal standard mix was spiked (10 μL of a 0.2 μg L−1 solution) onto the external surface of each PE tube including a procedural laboratory blank. MeOH (4 mL) was pipetted into each Falcon tube and tubes were sonicated for 10 min.. The extract was transferred into a clean, pre-labelled Falcon tube and fresh MeOH (4 mL) was added to the original tube containing the PE sampler for an additional 10 min. of sonication. The process was repeated another two times. The combined extracts were reduced to 1 mL under a stream of nitrogen and then centrifuged (3500 rpm for 10 min @ 22 °C). The supernatant was transferred into 1.5 mL plastic LC vials and evaporated to 200 μl. Samples were made up to a final volume of 500 μL (MeOH: MQ 40:60 v/v) and spiked with 10 μL of the M8PFOS and M8PFOA recovery standard mixture. Sample vials were placed in the refrigerator (4 °C) until analysis (within one week). The protocol for extraction of grab samples is detailed in the SI.
Cs = Ksw Cw 1
ln(2) ms Ksw Rs
(1) (2) −1
where CS is the concentration of the compound in the sampler (ng g ), KSW is the sorbent-water partition coefficient (L kg−1), CW the 425
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
Fig. 1. Uptake curves showing mass of PFASs accumulated (ng PE−1) in PE passive sampler vs. time (days−) for each of the five groundwater deployment sites (Sites 1–5). The bar graph inserts represent the calculated sampling rates (Rs) for each PFASs (where available) across four sampling sites (Sites 1–4).
426
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
Table 1 Relevant PFASs chemical properties, and sampling rates (Rs: mL d−1) (with standard deviations, ± SD and r2) estimated at each of the five in situ groundwater calibration sites. PFAS
PFBA PFPeA PFHxA PFHpA PFOA PFNA PFDA PFBS PFHxS PFOS 6:2 FTS 8:2 FTS
Molecular formula
C4HF7O2 C7HF9O2 C6HF11O2 C7HF13O2 C8HF15O2 C9HF17O2 C10HF19O2 C4HF9SO3 C6HF13O3S C8HF17SO3 C8H5F13O3S C10H5F17O3S
Molar Mass
Site 1
Site 2
Site 3
Site 4
Site 5*
(g mol−1)
Rs (mL d−1)
± SD
r2
Rs (mL d−1)
± SD
r2
Rs (mL d−1)
± SD
r2
Rs (mL d−1)
± SD
r2
Rs (mL d−1)
± SD
r2
214 264 314 364 414 464 514 300 400 500 428 528
1.68 2.11 2.18 2.33 2.45
0.17 0.15 0.27 0.27 0.34
0.94 0.92 0.86 0.87 0.83
1.79 2.46 2.57 2.69 2.67
0.12 0.27 0.15 0.18 0.08
0.92 0.94 0.94 0.94 0.98
2.24 3.37 3.98 3.48 3.3
0.1 0.28 0.18 0.07 0.15
1.0 1.0 1.0 1.0 1.0
2.1 2.69 3.06 3.01 3.07 3.62
0.1 0.11 0.1 0.11 0.15 0.2
0.96 0.98 0.98 0.98 0.96 0.95
2.37 2.41 4.3*
0.35 0.38 0.66*
0.85 0.75 0.99*
2.9 3.27 4.37
0.2 0.21 0.1
0.93 0.94 0.99
3.88 3.6 4.76 2.6
0.1 0.11 0.26 0.13
1.0 1.0 1.0 1.0
3.34 3.89 4.97 2.96 4.19
0.18 0.23 0.42 0.17 0.66
0.95 0.93 0.89 0.93 0.64
0.31 0.37 0.85 0.74 0.63 0.66 1.3 0.54 2.01 1.02 0.61 0.70
0.02 0.05 0.07 0.07 0.05 0.03 0.06 0.05 0.1 0.03 0.02 0.17
0.98 0.95 0.98 0.98 0.98 0.99 0.99 0.98 0.99 0.99 0.99 0.89
Average Rs (n = 4) (mL d−1)
± SD
%CV
1.95 2.66 2.95 2.88 2.87
0.26 0.53 0.77 0.49 0.38
13% 20% 26% 17% 13%
3.12 3.29 4.7 2.78
0.64 0.64 0.31 0.26
20% 19% 7% 9%
a = no significant adsorption. N/A = not measured. *The last data time point (29/08/17) in the calibration for this analyte/site was excluded. Perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluorobutane sulfonate (PFBS), perfluorohexane sulfonate (PFHxS), perfluorooctane sulfonate (PFOS), 6:2 fluorotelomer sulfonate (6:2 FTS), 8:2 fluorotelomer sulfonate (8:2 FTS).
3.2. Uptake of PFASs in PE passive samplers
yields an estimated mean t1/2 for PFASs in groundwater using the PE sampler of 240 days (with a range of 122–490 days for PFOS and PFOA, respectively, noting that modelled estimates using available data were only possible for PFHpA, PFHxA, PFOS, PFBS, PFOS, Table S5). Since the time required for the sampler to reach equilibrium according to Eq. (1) is strictly infinite, we define an effective equilibrium time (teq) as the finite time for Cs to reach 99% of its equilibrium value, i.e. teq = 4.605msKsw/Rs. Using the above data for ms, Ksw and Rs we estimate 1611 days (or 4.4 years) to achieve equilibrium with 810 and 3270 days for PFOS and PFOA, respectively. Hence, in comparison with the POCIS that reported t1/2 of up to only 13 days and teq of up to 60 days, the linear range for the sampling of PFASs was significantly extended. The configuration and composition of the PE sampler was successful at slowing down the uptake of PFASs. The increase in the linear range of sampling means that samplers can be deployed to accumulate PFASs for extended periods that can be extremely beneficial given the persistence and stability of these analytes in water systems. The calculated LODs for PFASs with the PE sampler ranged from 0.003 – 0.001 ng mL−1 (based on sampling rates of 2-5 mL d−1 over 15 days and an instrumental average LOD of 0.1 μg mL−1) representing adequate sensitivity.
Uptake of PFASs in PE passive samplers was observed for the five sites investigated (i.e. Sites 1–5, Fig. 1). A different uptake trend was observed for Site 5, therefore the discussion immediately following is in reference to the results observed from Sites 1–4, while Site 5 is discussed separately and in more detail, including an examination of limitations, further on. Linear accumulation of twelve PFASs (r2 ≥ 0.84 for most PFASs; Table 1) in the PE passive samplers was observed over the 82 or 83 day deployment periods (Eq. (2) and Fig. 1). The masses of PFASs accumulated in PEs ranged from 0.0003 μg sampler−1 for PFDA (at day 14) to 1.6 μg sampler−1 for PFHxS (at day 82). As indicated by the grab sampling data, mass of PFASs accumulated was highest at Sites 4 and 5, followed by Site 3 > Site 1 > and Site 2. The predominant PFASs detected in samplers across the sites were PFOS and PFHxS followed by PFOA, consistent with the grab sampling data. In some cases, linear accumulation of PFASs (i.e. PFNA, PFDA, 6:2 FTS and 8:2 FTS) (Fig. 1) was observed in PE samplers where there was no detection in the grab samples due to the relatively low levels of these analytes in groundwater. PFNA linearly accumulated in PE samplers at four sites: Sites 1–4 (Fig. 1), but was only reportable above LOD in grab samples at two sites: Sites 4 and 5 (sites with the highest ΣPFAS concentration). Similarly, PFDA was below LOD in all grab water samples but was linearly accumulated in PE samplers at Sites 4 and 5. This highlights the benefit of the PE passive samplers at the lower range of detection limits where enrichment from water over a two-week period was sufficient to reach reporting capabilities.
3.4. Sampling rates of PFASs in PE passive samplers Sampling rates (Rs) for PFASs at Sites 1–4 were estimated from regression analysis of the linear uptake data (Eq. (2)) and ranged from 1.68 to 4.97 mL d−1 for PFBA and PFOS, respectively (Table 1). Sampling rate data for individual PFASs from Sites 1 to 4 exhibited little variance (%CV = < 26%, Table 1). Consequently, relative sampling rates were consistent across these sites. For example, Rs for PFOA 1.6 ± 0.12 times lower than for PFOS, which consistently had the highest sampling rate (4.7 ± 0.3 mL d−1). As stated previously, despite accumulation of some PFASs (e.g. PFNA, PFDA, 6:2 FTS and 8:2 FTS) in the passive sampler, Rs could not be calculated as levels were below the LOD in grab samples. The mean Rs for PFAS in this study (approximately 3 mL d−1) is some 90 times lower than that estimated for the modified POCIS passive sampler (approximately 270 mL d−1) despite a relatively small difference in the sampler’s surface area (10 cm2 and 16 cm2, respectively) [15]. A reduced Rs is desirable for a PFAS sampler in groundwater
3.3. Estimation of the duration of the linear uptake mode of the PE sampler An important factor for the applicability of the sampler is its effective deployment duration. For a sampler in linear uptake or kinetic mode, this is the half-time (t1/2) i.e. the time it takes for Cs (PFASs) to be half its equilibrium value of KswCw. Linear uptake mode of PFASs is desirable as it is only possible to calculate time-weighted-average water concentration estimates from passive samplers within this sampling phase window. To derive t1/2 (Eq. (3)) the mass of sorbent (Strata XAW) in samplers is 4 × 10−4 kg, mean Ksw 2642 L kg-1 [15]) and the mean sampling rate (Rs) as determined in this study 0.003 L d-1. This 427
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
sorption coefficient and ks the PFASs mass transfer coefficient. Alvarez et al. adopt a similar approach for POCIS in describing trans-membrane flux of diffusants as a weighted average of transport through both the polymer matrix and aqueous pores [27]. With the POCIS, a decrease in Rs of a factor of 1.5 with molar mass for the same suite of PFASs as employed in the current work was observed. This was taken as evidence that uptake was WBL controlled. However the present direct linear relationship (Fig. 2) implies a more complex situation that may involve the PE as well as aqueous barriers to mass transfer. The relatively small Rs values determined with the PE sampler are of great advantage in groundwater applications, as this means minimal analyte depletion in the system. In most groundwater situations, where significant flow or turbulence is unlikely, depletion of the water in the vicinity of the sampler due to essentially stagnant conditions may become a problem. Although the reproducibility of Rs data between sampling sites in this study is promising and may suggest the influence of reduced flow rates, we were not able to deploy monitors in the narrow groundwater bores to characterise local flow rates. Further investigations under a range of different flow conditions would further validate the diffusion characteristics and flow effects on PFASs sampling rates when using this PE passive sampler.
Fig. 2. Plot of sampling rate (Rs, mL d−1) versus molar mass (g mol−1) for those PFASs where parallel grab sampling data was available.
environments as it contributes to a longer linear uptake phase and extended monitoring capabilities. With the present PE sampler, a direct linear relationship (r2 = 0.92) of Rs with PFASs molar mass was evident, such that an increase in Rs by a factor 1.6 was observed over the PFASs molar mass range of 314–514 g mol−1 (Fig. 2). As well as facilitating the estimation of sampling rates for uncalibrated PFASs, this relationship may provide insight into the mass transfer process operating under the conditions at the groundwater sites. If the accumulation of PFASs can be considered as a multi-stage process through a water boundary layer (WBL) and PE matrix or aqueous pores in the PE to the sorbent, the overall mass transfer coefficient (ko = Rs/A, where A is the surface area of the sampler) can be represented by [15,24]
1 A = = w + ko Rs Dw
p
Dw + (1
2
) Dp Kpw
2
+
1 Ksw ks
3.5. Atypical sampling behaviour and recognition of limitations The PE samplers at Site 5 exhibited substantially lower sampling rates (by a factor of approximately 4) when compared to all other sites (Table 1). While concentrations of the PFASs determined from the grab samples appear to be more variable at this site, this variability cannot readily explain the reduced sampling rates observed. It is noted that uptake rates appear linear during the first four sampling periods with a noticeable depletion of the sampler towards the last sampling point (i.e. 29/08/17). The coefficient of variation for derived water concentrations between triplicate PE samplers at this last sampling time point were considerably higher (%CV = 39%) in comparison with the average of all other PE sampler triplicates (%CV = 14%). The characteristics of this groundwater bore were notably different, being characterised by distinct hydrocarbon odour (likely due to its central location and adjacent activities). It was also noted during the penultimate sampling period that some training activity was conducted adjacent to the borehole that used an unknown quantity of fresh water on the surface. Currently, we have no suitable explanation for the atypical sampling behaviour at this site. It is possible that the high concentration of hydrocarbons at Site 5 has influenced uptake, however the
(4)
where the first term represents resistance to mass transfer in the WBL (δw is WBL thickness and Dw the PFASs diffusion coefficient in water). The second term describes parallel resistance to PFASs transfer in PE and aqueous pores with δp (2 mm) being PE thickness, Dp the PFASs diffusion coefficient in PE, Kpw the PE/water distribution coefficient of the PFASs and θ and the PE tortuosity and porosity respectively. The final term represents the sorbent phase. In this, Ksw is the PFASs
Fig. 3. Mean concentrations of PFASs in groundwater from PE samplers (ng L−1; n = 3) vs Grab samples (n = 1) at all sites, for all PFASs. The dotted red line represents a 1:1 ratio (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
428
429
Grab Sample
37 38 48 38 48 18 5.7 8.4 11 179 39 997 7.6 5.4 6.2 < LOD 1.1
Ave
0.4
3.9 6.4 9.4 4.9 5.0 1.7 0.3 3.2 4.1 52 16 173 2.1 0.9 1.2
± SD
117 150 181 139 270 73 19 31 40 660 66 4865 24 40 20 < LOD < LOD
38 47 62 53 71 28 6.6 14 21 337 67 848 2.3 3.3 1.2 < LOD 0.3
Ave
0.1
1.4 1.2 1.6 5.7 7.3 4.9 0.8 0.5 3.1 95 7.4 153 0.7 0.8 0.2
± SD
PE Samplers (n = 3)
118 223 1084 124 88 0.1 < LOD 684 775 2619 8 N/A < LOD < LOD 0.3 < LOD < LOD
PE Samplers (n = 3)
< LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD
Ave
Site E
0.16 0.01 0.09
0.29
± SD
Site D
0.39 < LOD 0.17 0.06 0.24 < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD
Ave
72 105 142 89 168 27 < LOD 41 44 703 56 3816 < LOD < LOD < LOD < LOD < LOD
Grab Sample
0.1
6 17 70 8 9 0.03 < LOD 31 22 161 1.5 3
± SD
PE Samplers (n = 3)
PE Samplers (n = 4)
Grab Sample
Site A
Blanks
1241 2894 11270 3099 6203 96 2.0 4208 4808 36399 2668 61320 26 1.8 317 2.3 1.2
Ave
359 763 347 384 526 117 < LOD 73 63 315 36 319 < LOD < LOD 1.2 < LOD < LOD
Ave
0.2
15 23 12 13 90 12 < LOD 4.0 7.6 50 3.4 5.1
± SD
357 532 259 271 464 77 < LOD 47 38 302 25 550 < LOD < LOD < LOD < LOD < LOD
Grab Sample
159 222 876 286 684 26 0.3 295 511 5816 314 9311 3.9 0.3 28 0.1 0.1
± SD
PE Samplers (n = 3)
PE Samplers (n = 3)
Site F
119 191 1207 120 180 < LOD < LOD 926 682 2711 14 1160 < LOD < LOD < LOD < LOD < LOD
Grab Sample
Site B
154 272 102 199 272 104 < LOD 12 9.1 78 35 286 < LOD < LOD < LOD < LOD < LOD
Ave
14 31 8.4 29 18 7.2 < LOD 1.8 1.1 5.0 4.3 19
± SD
PE Samplers (n = 3)
Site G
2824 3768 13057 2499 5433 58 < LOD 5091 6186 34178 4995 49144 < LOD 30 325 < LOD < LOD
Grab Sample
199 250 85 124 262 64 < LOD < LOD < LOD 77 21 561 < LOD < LOD < LOD < LOD < LOD
Grab Sample
80 129 181 130 154 69 23 40 64 700 127 4015 24 29 40 < LOD 0.7
Ave
0.04 0.04 0.03 0.01 0.04 0.03 0.02 0.03 0.03 0.04 0.02 0.03 0.02 0.06 0.06 0.23 0.05
LOD (ng mL−1)
0.8
2.9 9.4 23 19 37 12 3.5 6.8 11 101 0.7 869 23 6.9 6.9
± SD
PE Samplers (n = 3)
Site C
0.14 0.13 0.11 0.04 0.12 0.09 0.06 0.11 0.10 0.13 0.07 0.10 0.07 0.18 0.19 0.77 0.17
LOQ (ng mL−1)
123 153 211 134 293 67 15 49 55 856 99 6240 < LOD 69 29 < LOD < LOD
Grab Sample
< LOD - below limit of detection. N/A - could not be quantified due to chromatographic interference. perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluorobutane sulfonate (PFBS), perfluoropentane sulphonate (PFPeS), perfluorohexane sulfonate (PFHxS), perfluoroheptane sulphonate (PFHpS), perfluorooctane sulfonate (PFOS), perfluorononane sulfonate (PFNS), 8:2 fluorotelomer sulfonate (8:2 FTS), 6:2 fluorotelomer sulfonate (6:2 FTS), 4:2 fluorotelomer sulfonate (4:2 FTS), perfluorooctane sulfonamide (FOSA).
PFBA PFPeA PFHxA PFHpA PFOA PFNA PFDA PFBS PFPeS PFHxS PFHpS PFOS PFNS 8:2 FTS 6:2 FTS 4:2 FTS FOSA
PFASs (ng L−1)
PFBA PFPeA PFHxA PFHpA PFOA PFNA PFDA PFBS PFPeS PFHxS PFHpS PFOS PFNS 8:2 FTS 6:2 FTS 4:2 FTS FOSA
PFASs (ng L−1)
Table 2 Comparison of water concentration estimates of 17 PFASs from PE passive samplers and grab samples (including limits of detection (LODs) and quantification (LOQ)), from a validation study conducted adjacent to a fire training ground, between 27.03.2018 - 28.6.2018.
S.L. Kaserzon et al.
Journal of Hazardous Materials 366 (2019) 423–431
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
mechanisms that influence this interaction are currently poorly understood [28].
Acknowledgements QAEHS is a joint venture of The University of Queensland and Queensland Health and Forensic Scientific Services (QHFSS). We would like to acknowledge Ben Tscharke, Robert Gray, Kristie Thompson, Gabriele Elisei, Sean Van Niekerk and Jake O’Brien for assistance with sampler deployment and retrieval. Part of this study was supported by Australian Research Council (ARC) Linkage grant LP160100510.
3.6. Validation trial of PE samplers and comparison with grab samples A field deployment was conducted in order to validate PE sampler performance by using the previously determined calibration data for a different groundwater contamination scenario and comparing samplerderived concentrations with those from grab sampling. PE samplers were deployed at 7 groundwater sites that receive various inputs from AFFF activities (Table S2). Good correlation was observed when plotting data from all sites and PFASs investigated (r2 = 0.98, Fig. 3), even across the wide range of contamination levels observed between the sites. For example PFOS and PFHxS concentrations ranged from 0.3 and 0.1 ng mL−1 (Site G) to 60 and 36 ng mL−1 (Site B), respectively (Table 2). Fifteen PFASs were detected in grab samples, while 17 were detected in PE passive samplers. In particular, the PE sampler was more effective at detecting precursor PFASs (i.e. 4:2 FTS, and FOSA present at relatively low levels (< 2.3 ng L−1). Mean sampling rates from the PE sampler calibration study (i.e. Table 1) were used to estimate PFASs water concentrations at the field deployment site (based on Eq. (2)). Where sampling rate data was not available (i.e. for PFNA, PFDA, PFPeS, PFHpS, PFNS, 8:2 FTS, 4:2 FTS and FOSA), modelled sampling rates were used based on the sampling rate (Rs, L d−1) versus molar mass (g mol−1) relationship depicted in Fig. 2. Focussing on data from individual sites, a linear correlation between PE samplers and grab sample derived concentrations at individual sites of with r2 values of > 0.96 was observed for five sites, and 0.68 and 0.76 observed for Sites G and F respectively. This decreased agreement in the latter two cases case may be the result of the close vicinity of these two adjacent wells to a marine embayment and consequent tidal influence. Depending on when a single point-in-time grab sample is collected within the tidal cycle may cause variability. It should be remembered that the PE samplers provide a time-weighted average water concentration estimate over 93 days, while the grab samples represent a single point in time viz. concentrations at sampler retrieval. Despite this, overall, a close correlation between PFASs concentrations with the two methods was apparent.
Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jhazmat.2018.12.010. References [1] Z. Wang, J.C. Dewitt, C.P. Higgins, I.T. Cousins, A Never-Ending Story of Per- and Polyfluoroalkyl Substances (PFASs)? Environ. Sci. Technol. 51 (2017) 2508–2518. [2] R.C. Buck, P.M. Murphy, M. Pabon, Chemistry, properties, and uses of commercial fluorinated surfactants, in: P.T. Knepper, T.F. Lange (Eds.), Polyfluorinated Chemicals and Transformation Products, Springer, Berlin Heidelberg, Berlin, Heidelberg, 2012, pp. 1–24. [3] J. Bräunig, C. Baduel, A. Heffernan, A. Rotander, E. Donaldson, J.F. Mueller, Fate and redistribution of perfluoroalkyl acids through AFFF-impacted groundwater, Sci. Total Environ. 596-597 (2017) 360–368. [4] E.F. Houtz, C.P. Higgins, J.A. Field, D.L. Sedlak, Persistence of perfluoroalkyl acid precursors in AFFF-Impacted groundwater and soil, Environ. Sci. Technol. 47 (2013) 8187–8195. [5] A. Kärrman, K. Elgh-Dalgren, C. Lafossas, T. Møskeland, Environmental levels and distribution of structural isomers of perfluoroalkyl acids after aqueous fire-fighting foam (AFFF) contamination, Environ. Chem. 8 (2011) 372–380. [6] L. Ahrens, Polyfluoroalkyl compounds in the aquatic environment: a review of their occurrence and fate, J. Environ. Monit. 13 (2011) 20–31. [7] I.T.C.K. Prevedouros, R.C. Buck, S.H. Korzeniowski, Sources, fate and transport of Perfluorocarboxylates, Environ. Sci. Technol. 40 (2006). [8] T.M.Jennifer L. Guelfo, David M. Klein, David A. Savitz, Scott Frickel, Michelle Crimi, Eric M. Suuberg, Evaluation and management strategies for perand polyfluoroalkyl substances (PFASs) in drinking water aquifers: perspectives from impacted U.S. Northeast communities, Environ. Health Perspect. 126 (2018). [9] C. Ort, M.G. Lawrence, J. Rieckermann, A. Joss, Sampling for pharmaceuticals and personal care products (PPCPs) and illicit drugs in wastewater systems: are your conclusions valid? A critical review, Environ. Sci. Technol. 44 (2010) 6024–6035. [10] C. Baduel, C.J. Paxman, J.F. Mueller, Perfluoroalkyl substances in a firefighting training ground (FTG), distribution and potential future release, J. Hazard. Mater. 296 (2015) 56-53. [11] D. Szabo, T.L. Coggan, T.C. Robson, M. Currell, B.O. Clarke, Investigating recycled water use as a diffuse source of per- and polyfluoroalkyl substances (PFASs) to groundwater in Melbourne, Australia, Sci. Total Environ. 644 (2018) 1409–1417. [12] ISO Water quality, Sampling Part 23: Guidance on Passive Sampling in Surface Waters, ISO 5667-23:2011, ISO, Genève, Switzerland, 2016 in. [13] C. Harman, I.J. Allan, E.L.M. Vermeirssen, Calibration and use of the polar organic chemical integrative sampler-a critical review, Environ. Toxicol. Chem. 31 (2012) 2724–2738. [14] B. Vrana, I.J. Allan, R. Greenwood, G.A. Mills, E. Dominiak, K. Svensson, J. Knutsson, G. Morrison, Passive sampling techniques for monitoring pollutants in water, TrAC Trends Anal. Chem. 24 (2005) 845–868. [15] S.L. Kaserzon, K. Kennedy, D.W. Hawker, J. Thompson, S. Carter, A.C. Roach, K. Booij, J.F. Mueller, Development and calibration of a passive sampler for perfluorinated alkyl carboxylates and sulfonates in water, Environ. Sci. Technol. 46 (2012) 4985–4993. [16] S.L. Kaserzon, E.L. Vermeirssen, D.W. Hawker, K. Kennedy, C. Bentley, J. Thompson, K. Booij, J.F. Mueller, Passive sampling of perfluorinated chemicals in water: flow rate effects on chemical uptake, Environ. Pollut. 177 (2013) 58–63. [17] S.L. Kaserzon, D.W. Hawker, K. Booij, D.S. O’Brien, K. Kennedy, E.L. Vermeirssen, J.F. Mueller, Passive sampling of perfluorinated chemicals in water: in-situ calibration, Environ. Pollut. 186 (2014) 98–103. [18] K. Booij, N.L. Maarsen, M. Theeuwen, R. Van Bommel, Method to account for the effect of hydrodynamics on polar organic compound uptake by passive samplers, Environ. Toxicol. Chem. 36 (2017) 1517–1524. [19] J.K. Challis, M.L. Hanson, C.S. Wong, Development and calibration of an organicdiffusive gradients in thin films aquatic passive sampler for a diverse suite of polar organic contaminants, Anal. Chem. 88 (2016) 10583–10591. [20] D. Guan, Y. Li, N. Yu, G. Yu, S. Wei, H. Zhang, W. Davison, X. Cui, L. Ma, J. Luo, In situ measurement of perfluoroalkyl substances in aquatic systems using diffusive gradients in thin-films technique, Water Res. 144 (2018). [21] S. Bopp, H. Weiß, K. Schirmer, Time-integrated monitoring of polycyclic aromatic hydrocarbons (PAHs) in groundwater using the Ceramic Dosimeter passive sampling device, J. Chromatogr. A 1072 (2005) 137–147. [22] J. Cristale, A. Katsoyiannis, C. Chen, K.C. Jones, S. Lacorte, Assessment of flame retardants in river water using a ceramic dosimeter passive sampler, Environ. Pollut. 172 (2013) 163–169.
4. Conclusion The investigated microporous PE diffusion passive sampler represent the first passive sampling solution suitable for quantitative long-term monitoring of PFASs in groundwater. In calibration studies, twelve PFASs were detected in-situ in PE samplers, with linear accumulation observed over a three-month deployment period. Sampling rates (Rs) determined for PFASs (mean of approximately 3 mL d−1) were consistent at four of the five calibration sites. A field validation conducted at seven contaminated groundwater sites, using the previously determined calibration parameters, showed successful quantification of PFASs over a wide range of concentrations, and with generally good agreement with grab sampling data. Successful quantification of precursor PFASs (i.e. 4:2 FTS and FOSA) was also possible with the PE samplers, despite concentrations in grab samples being < LOD. The sampler’s relatively simple, robust and compact design facilitates deployment in narrow groundwater bores or other confined systems. In comparison with previous types of samplers, the duration of the linear uptake phase of PFASs in this design was successfully extended, as evidenced by a mean t1/2 estimate of 240 days. This makes the sampler useful for longer-term deployments (3 months or more) and for temporal as well as spatial assessments of PFASs groundwater concentrations. However, further investigation of the sampler’s performance in longer deployments of 6–12 months duration, under different physicochemical conditions and in additional environmental matrices, such as surface and wastewaters, would be beneficial to establish its wider applicability. 430
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al. [23] H. Franquet-Griell, V. Pueyo, J. Silva, V.M. Orera, S. Lacorte, Development of a macroporous ceramic passive sampler for the monitoring of cytostatic drugs in water, Chemosphere 182 (2017) 681–690. [24] V. Fauvelle, N. Montero, J.F. Mueller, A. Banks, N. Mazzella, S.L. Kaserzon, Glyphosate and AMPA passive sampling in freshwater using a microporous polyethylene diffusion sampler, Chemosphere 188 (2017) 241–248. [25] J.N. Huckins, J.D. Petty, K. Booij, Monitors of Organic Chemicals in the Environment: Semipermeable Membrane Devices, Springer Science & Business Media, 2006. [26] C. Baduel, J.F. Mueller, A. Rotander, J. Corfield, M.J. Gomez-Ramos, Discovery of novel per- and polyfluoroalkyl substances (PFASs) at a fire fighting training ground
and preliminary investigation of their fate and mobility, Chemosphere 185 (2017) 1030–1038. [27] D.A. Alvarez, J.N. Huckins, J.D. Petty, T.L. Jones-Lepp, F. Stuer-Lauridsen, D.T. Getting, J.P. Goddard, A. Gravell, Tool for monitoring hydrophilic contaminants in water: polar organic chemical integrative sampler (POCIS), in: R. Greenwood, G. Mills, B. Vrana (Eds.), Passive Sampling Techniques in Environmental Monitoring, Comprehensive Analytical Chemistry, 48, Elsevier, Amsterdam, 2007, pp. 171–197. [28] J.L. Guelfo, C.P. Higgins, Subsurface transport potential of perfluoroalkyl acids at aqueous film-forming foam (AFFF)-impacted sites, Environ. Sci. Technol. 47 (2013) 4164–4171.
431
Contents lists available at ScienceDirect
Journal of Hazardous Materials journal homepage: www.elsevier.com/locate/jhazmat
Calibration and validation of a novel passive sampling device for the time integrative monitoring of per- and polyfluoroalkyl substances (PFASs) and precursors in contaminated groundwater
T
Sarit L. Kaserzona, , Soumini Vijayasarathya, Jennifer Bräuniga, Linus Muellera, Darryl W. Hawkerb, Kevin V. Thomasa, Jochen F. Muellera ⁎
a b
Queensland Alliance for Environmental Health Sciences (QAEHS), The University of Queensland, 20 Cornwall street, Woolloongabba, Queensland, 4102, Australia Griffith School of Environment and Science, Griffith University, Nathan, Queensland, 4111, Australia
ARTICLE INFO
ABSTRACT
Keywords: Per-and polyfluoroalkyl substances (PFASs) Passive sampling Groundwater Microporous polyethylene Sampling rate
Per-and polyfluoroalkyl substances (PFASs) as key components in aqueous film forming foams (AFFF) have led to growing incidences of environmental contamination. The aim of this study was to investigate a novel diffusion based passive sampling device comprising of microporous polyethylene (PE) for the long-term time-integrative monitoring of PFASs in groundwater systems. PE passive samplers (PEs) were deployed for 83 d and calibrated at five AFFF impacted groundwater sites representing different PFASs concentration levels (ΣPFAS 0.001 to 0.1 ng mL−1). Grab samples were collected simultaneously. Linear accumulation of 12 PFASs (r2 ≥ 0.84) were observed in the PEs over 83 d and PFASs sampling rates were 2–5 mL d−1. Estimated mean half-times to equilibrium for PFASs ranged between 122 and 490 d. A separate validation study compared PEs and grab sampling during a 93 d field deployment, at seven groundwater sites near a fire fighting training ground. Seventeen PFASs were detected in PEs and fifteen in grab samples. PEs showed higher sensitivity for precursors (i.e. 4:2 FTS and FOSA). Time-weighted-average water concentrations across all validation sites for all PFASs determined from PEs were strongly correlated (r2 = 0.98) with grab samples, (within range 0.3–60 ng mL-1 PFOS). Results represent the first application of passive sampling technology for the quantitative assessment of PFASs in groundwater systems.
1. Introduction Per- and polyfluoroalkyl substances (PFASs) comprising compounds such as perfluoroalkylcarboxylates (PFCAs; CnF2n+1COO−) and sulfonates (PFSAs; CnF2n+1SO3−) have become ubiquitous and persistent environmental trace contaminants with over 3000 estimated to be used in various formulations on the global market [1,2]. Among the various industrial and domestic uses of these chemicals, it is estimated that approximately 200–300 occur in aqueous film-forming foams (AFFF) [1,3–5]. As a result of the physico-chemical properties such as aqueous solubility of some of the most widely distributed PFASs (i.e. PFOS, PFOA and PFHxS), aqueous phases play an important role in the environmental transport and fate of these chemicals [3,5–8]. A recent U.S study has identified that the main source of PFASs and highest risk factor for contamination in drinking water aquifers results from AFFF use (Department of Defense (DoD)) followed by PFASs manufacturing, waste streams i.e. landfills, AFFF use (airports) and fire-training areas
⁎
[8], emphasising the need for a diverse range of PFASs monitoring and management strategies. The collection of representative samples is a requirement for estimating PFASs concentrations over defined time scales and for accurate assessment of environmental burden [9]. In surface and wastewaters, the occurrence and extent of episodic events such as accidental spills or deliberate industrial release can be challenging to capture and adequately quantify. Contaminant levels in groundwater may show orders of magnitude difference between sampling sites, depending on factors such as hydraulic and geological conditions of the area, and the paths of chemical transport [3,8]. Leaching of PFASs into the environment from primary sources can persist for many decades [10,11]. However there is still a lack of understanding of what influences the mobility of PFASs in soils and groundwater [10]. Passive sampling techniques [12] have the potential to provide insight to groundwater contamination and hence become important complementary tools in PFASs pollution monitoring schemes. Some of the highlighted advantages of these techniques in this
Corresponding author. E-mail address: [email protected] (S.L. Kaserzon).
https://doi.org/10.1016/j.jhazmat.2018.12.010 Received 9 September 2018; Received in revised form 13 November 2018; Accepted 2 December 2018 Available online 07 December 2018 0304-3894/ © 2018 Elsevier B.V. All rights reserved.
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
regard include the ability to provide time-weighted average water concentration estimates over extended periods (i.e. days to weeks), relatively low limits of detection, cost effectiveness as well as ease and flexibility of deployment [13,14]. In 2012, the first calibrated passive sampler for the monitoring of a range of PFASs in surface waters, based on a modified Polar Organic Chemical Integrative Sampler (POCIS), was developed [15–17]. Results showed that all PFASs targeted were successfully sampled. However, equilibration of PFASs in the sampler was relatively fast with half-times to equilibrium (t1/2) for PFOS and PFOA ranging from 2.4 to 13 days. This meant samplers could only be operated in time-integrative mode for a relatively short period of time. Furthermore, the mass transfer and accumulation of PFASs in the samplers (i.e. sampling rates of PFASs) were shown to be influenced by deployment-specific conditions specifically, water turbulence [16], which could in turn cause the over- or under-estimation of water concentrations if laboratory-derived sampling rate data were employed instead of in-situ values [18]. Recently, a diffusion gradient in thin film (DGT) based passive sampler design (surface area 3.1 cm2) for organic chemicals (o-DGT) has been investigated as an alternative design for the sampling of polar organic chemicals [19]. Further development of the DGT sampler with XAD18 resin was employed for the detection of perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) in surface and waste waters. The reported diffusion coefficients (298 K) in the diffusive gel for this sampler were 4.37 × 10 − 6 and 5.08 × 10 − 6 cm2 s−1 (PFOA and PFOS, respectively). This, together with relatively high capacity for these compounds, suggested deployment times of several weeks were possible [20]. The use of a diffusion gel allows for a more accurate estimation of the mass transfer coefficient (i.e. sampling kinetics) of pollutants into the sampler with minimal effect from water turbulence or flow velocity on the sampling rates of chemicals. This is because the water boundary layer (WBL) thickness is typically much less than the gel thickness, resulting in a reduced influence of the WBL. However, challenges with this sampler design include relatively small sampling window surface areas and consequently lower sampling rates, susceptibility of the diffusion gels to depletion and biodegradation in some environmental systems which result in relatively short recommended deployment periods (2–4 weeks), and samplers that may be too big to fit into narrow groundwater bores [19,20]. Bopp et al. (2005) proposed a design suitable for groundwater deployment comprising of 8 cm long, 2 mm thick ceramic dosimeter tubes filled with sorbent (Amberlite IRA-743) for the passive sampling of VOCs [21]. In an aim to eliminate any WBL effects, a microporous diffusion layer replaced the membrane associated with POCIS. These tubes were trialled for monitoring flame retardants in river water [22] and for monitoring of cytostatic drugs in water [23]. However, it has been recognised that in some environmental matrices the ceramic pores of the sampler may be susceptible to erosion and/or clogging [23]. Furthermore the use of ceramic material for PFASs may be problematic, as PFASs demonstrate some affinity for silica (i.e. ceramics) negating its effectiveness as a suitable diffusion layer. Hence, sampling and laboratory material that are suitable for PFASs work often consist of polyethylene materials. In 2017, we successfully validated a similar tubular passive sampling device of the same dimensions manufactured from a cheap and robust microporous polyethylene (PE) enclosing a TiO2 receiving phase, for the monitoring of glyphosate and its transformation product aminomethylphosphonic acid (AMPA), both of which exist in ionized form in most surface waters [24]. The chemical diffusion characteristics and linear accumulation of such a sampler observed with these challenging anionic analytes raised the prospect of its successful application for PFASs, also environmental contaminants of contemporary concern. The overall aim of this study then was to develop a sensitive and robust monitoring tool for long term quantitative assessment of PFASs in contaminated groundwater. For this purpose, we investigate the microporous PE tube material in combination with an anion exchange
sorbent (Strata X-AW) as a passive sampling device. Firstly, the uptake kinetics and duration of the linear sampling mode were determined at five groundwater sites representing various PFASs contamination levels from historic AFFF activities. Any concentration dependency of the sampler and whether this may affect quantitative water concentration estimates was examined. We then apply the in-situ model calibration parameters obtained to estimate water concentrations of PFASs in groundwater sites located adjacent to a fire fighting training ground, and compare these to parallel grab sampling data. 2. Materials and methods 2.1. Microporous PE passive sampler calibration site description Microporous PE passive water samplers were deployed at five groundwater bores (denoted Sites 1–5) for periods of 14 or 15, 30 or 31, 40 or 41, 61 or 62 and 82 or 83 days. Site selection was based on (i) the ability to easily and safely approach and access sites (ii) a range of PFASs concentrations (i.e. low, medium and high) based on earlier studies. The contaminated sites sampled had restricted access and timepoints were chosen based on feasibility, relevance to users of the technology and access to sites. Site characteristics such as temperature, pH and electrical conductivity were collected at each sampling location throughout the sampling period (Table S1). 2.2. Microporous PE passive sampler validation site description The PE passive samplers were deployed at a total of 7 groundwater wells across 3 sites located adjacent to a fire training ground in Australia. Triplicate PE samplers were deployed for 93 days. Grab samples (500 mL) were collected at each site upon retrieval of the PE samplers from the wells. Information including depth of sampling, pH and salinity of water at sampling sites is included in Table S2. Grab samples (10 mL) from both validation and calibration sites were extracted as detailed by Bräunig et al. [3] (a brief description is provided in the Supporting Information (SI)). 2.3. Standards and reagents All equipment and materials (including glassware) were thoroughly rinsed with acetone (ACE) and methanol (MeOH) (Merck, Darmstadt, Germany; purity 99.8%) and allowed to dry prior to use. Ammonium acetate (> 97%, C2H7NO2) was obtained from ChemSupply (Gillman, SA, Australia). Ammonium hydroxide solution (NH4OH) was obtained from Sigma Aldrich. Water with resistivity > 18.2 MΩ cm (MQ) was obtained from a Millipore system. Samples were analysed for 29 PFASs and precursors (Detailed in the SI) including perfluoroalkyl sulfonates (PFSAs), perfluoroalkyl carboxylates (PFCAs), perfluorooctane sulfonamides (FOSAMs) and fluorotelomer sulfonates (FTSs). All analytical standards were purchased from Wellington Laboratories, Guelph, Canada. A native standard mix (PFAC-MXB, Wellington Laboratories) along with additional PFASs was used to prepare calibration standards in 40:60 MeOH:MQ (with 5 mM ammonium acetate) with a range from 0.1 to 100 ng mL−1. Mass labelled standards from Wellington Laboratories, MPFAC-MXA (13C4-PFBA, 13C2-PFHxA, 13C4-PFOA, 13C5PFNA, 13C2-PFDA, 13C2-PFUnDA, 13C2-PFDoDA, 18O2-PFHxS, 13C4PFOS) and M3PFBS (13C3-PFBS), M3PFPeA (13C3-PFPeA), M4PFHpA (13C4-PFHpA), M26:2 FTS (13C2-6:2 FTS) were used as surrogate internal standards. A mixture of M8PFOS and M8PFOA (13C8-PFOS and 13 C8-PFOA), were used as recovery standards. 2.4. Sample preparation and extraction 2.4.1. Passive sampler preparation Microporous polyethylene (PE) tubes (PALL, Germany) 0.8 cm in external diameter, 2 mm thick, with a pore size of 2.5 μm and 35% 424
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
concentration of the compound in water (ng L−1), RS the sampling rate (L day−1), t the time deployed (days) and mS the mass of the sorbent (kg). The half-life time to equilibrium (t1/2), describing the time it takes Cs to be half its equilibrium value is given by
porosity were cut into 4 cm lengths (Figure S1). Samplers were cleaned by first submerging them in a glass jar with acetone on a shaker for 24 h, followed by submergence in fresh acetone for another 24 h. This was followed by 2 x 24 h cleans in methanol. Samplers were left to dry in a fume hood. Each PE passive sampler was then packed with 400 ± 10 mg of sorbent phase material (Strata X-AW; Phenomenex, Australia). Both ends of the sampler were capped with 10 mm plastic round tubing insert caps (0.8–1.5 mm wall thickness; STOCKCAP, Australia). PE samplers were conditioned in methanol (1 x 24 h) on a shaker followed by 2 x 24 h in MQ water. All PE passive samplers were kept submerged in MQ water until deployment. Laboratory blank PE samplers were prepared alongside samplers (n = 6) and kept in the refrigerator until the processing and extraction of PE samplers.
t1/2 =
2.7. Quality control Laboratory blank passive samplers (n = 6) as well as laboratory and field blank grab samples (n = 7) were prepared, extracted and analysed in parallel with each batch of exposed samples. No PFASs were detected in blank PE samplers above LOQ while blank levels detected in grab samples are presented in Table S3. The method limit of detection was calculated as average blank level plus 3 times the standard deviation (SD). Instrument detection (LOD) and quantitation (LOQ) limits were calculated by multiplying the SD obtained from injecting the lowest calibration standard eight times by 3 and 10, respectively. Acceptable coefficients of variation (%CV) between triplicate PE passive samplers values were observed with < 28% for the majority of samples (Table S4). Grab samples showed %CV of ≤ 19% for intra-day grab samples, with the exception of 6:2FTS and 8:2FTS for which %CV was ≤ 60% due to trace levels amounts detected as reported in Table S3. Recovery of chemicals was verified by spiking blank and exposed samplers with PFASs surrogates prior to extraction, and recovery standards prior to analysis. Non-extracted side spikes (solvent blanks spiked with surrogates and recovery standards) were prepared in parallel to spiking and extracting exposed samples. These represent 100% recoveries and are essential in recovery correction calculations. Acceptable recoveries (typically between 42%–165%) were observed for PFASs passive samples (Table S4) with similar recoveries (30%–167%) for grab samples.
2.5. Analytical method All samples were analysed by high performance liquid chromatography tandem mass spectrometry (HPLC-MS/MS) using a Nexera HPLC (Shimadzu Corp., Kyoto, Japan) coupled to API5500 QTRAP mass spectrometer (Sciex, Melbourne, Australia) with an electrospray ionization (ESI) interface operating in negative ion mode. Chromatographic separation of the analytes was achieved with a Gemini C18 column (50 x 2.0 mm, 4 μm; Phenomenex, Torrance, CA), maintained at 45 °C, with a flow rate of 0.3 mL min−1 and injection volume of 5 μL. Mobile phases consisted of MeOH: water (1:99 v/v) (A), and MeOH: water (95:5 v/v) (B), with 5 mM ammonium acetate in both phases. An isolator column (Phenomenex) was included inline directly after the mobile phase mixing chamber to delay the elution of any solvent-derived background PFASs contamination. Data acquisition and processing was carried out using Analyst® TF 1.6 and MultiQuant™ software (Sciex).
3. Results and discussion 3.1. Concentrations of PFASs in groundwater samples Twelve PFASs were detected in grab samples across the five contaminated sites employed for calibration purposes, with eight consistently detected at all sites (Table S3). Mean PFASs groundwater concentrations (Cw) ranged from 0.01 ng mL−1 (for 6:2 FTS and 8:2 FTS) to 74 ng mL−1 for PFHxS. PFOS and PFHxS consistently exhibited the highest concentrations at all sites investigated followed by PFOA. ΣPFASs concentrations were 1.4 ng mL−1 (Site 1) < 3.7 ng mL−1 (Site 2) < 18.3 ng mL−1 (Site 3) < 26.8 ng mL−1 (Site 4) and < 157 ng mL−1 (Site 5) thus representing a range of contamination levels. 6:2 FTS and 8:2 FTS were detected only at sites with the highest ΣPFAS concentrations. Fluorotelomer sulfonates such as 6:2 FTS and 8:2 FTS have been reported to be degradation products of fluorotelomer-based AFFF foam (Ansulite), which replaced earlier AFFFs containing PFCAs/ PFSAs (3 M). A later transition to fluorine free foam (Solberg®) has been observed since 2010 [26]. No significant fluctuation of PFASs concentrations was observed between sampling days (%CV < 36% with the majority below 20% for all analytes and sites). Exceptions were found for 8:2 FTS at two sites (with %CV values of 49% and 60%) and PFHxS (with %CVs of 45% and 48%). Mean levels of these compounds represent concentration levels that were at the lower and upper range of the analytical calibration curve, respectively. Overall, concentrations of PFASs investigated were sufficiently stable to allow for the in-situ calibration of the PE passive sampling devices during the deployment period.
2.6. Data modelling Assuming a constant ambient aqueous phase concentration, the accumulation of ionizable and polar organic chemicals in the sorbent of polar passive samplers can be described using a one compartment, 1st order kinetic model (Equation 1). When uptake of analytes into the sampler is linear with time i.e. the sample is operating in kinetic mode, Equation 1 can be reduced to linear approximation model (Eq. (2)) [25].
Cs =
Cw Rs t ms
exp
Rs t ms Ksw
(3)
Values of Rs were estimated from Eq. (2), (by unweighted linear regression using GraphPad Prism 7.03). Ksw values were taken from Kaserzon et al. [15].
2.4.2. Passive sampler extraction Passive PE samplers retrieved from deployment were transferred into 15 mL Falcon tubes. The surrogate PFASs internal standard mix was spiked (10 μL of a 0.2 μg L−1 solution) onto the external surface of each PE tube including a procedural laboratory blank. MeOH (4 mL) was pipetted into each Falcon tube and tubes were sonicated for 10 min.. The extract was transferred into a clean, pre-labelled Falcon tube and fresh MeOH (4 mL) was added to the original tube containing the PE sampler for an additional 10 min. of sonication. The process was repeated another two times. The combined extracts were reduced to 1 mL under a stream of nitrogen and then centrifuged (3500 rpm for 10 min @ 22 °C). The supernatant was transferred into 1.5 mL plastic LC vials and evaporated to 200 μl. Samples were made up to a final volume of 500 μL (MeOH: MQ 40:60 v/v) and spiked with 10 μL of the M8PFOS and M8PFOA recovery standard mixture. Sample vials were placed in the refrigerator (4 °C) until analysis (within one week). The protocol for extraction of grab samples is detailed in the SI.
Cs = Ksw Cw 1
ln(2) ms Ksw Rs
(1) (2) −1
where CS is the concentration of the compound in the sampler (ng g ), KSW is the sorbent-water partition coefficient (L kg−1), CW the 425
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
Fig. 1. Uptake curves showing mass of PFASs accumulated (ng PE−1) in PE passive sampler vs. time (days−) for each of the five groundwater deployment sites (Sites 1–5). The bar graph inserts represent the calculated sampling rates (Rs) for each PFASs (where available) across four sampling sites (Sites 1–4).
426
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
Table 1 Relevant PFASs chemical properties, and sampling rates (Rs: mL d−1) (with standard deviations, ± SD and r2) estimated at each of the five in situ groundwater calibration sites. PFAS
PFBA PFPeA PFHxA PFHpA PFOA PFNA PFDA PFBS PFHxS PFOS 6:2 FTS 8:2 FTS
Molecular formula
C4HF7O2 C7HF9O2 C6HF11O2 C7HF13O2 C8HF15O2 C9HF17O2 C10HF19O2 C4HF9SO3 C6HF13O3S C8HF17SO3 C8H5F13O3S C10H5F17O3S
Molar Mass
Site 1
Site 2
Site 3
Site 4
Site 5*
(g mol−1)
Rs (mL d−1)
± SD
r2
Rs (mL d−1)
± SD
r2
Rs (mL d−1)
± SD
r2
Rs (mL d−1)
± SD
r2
Rs (mL d−1)
± SD
r2
214 264 314 364 414 464 514 300 400 500 428 528
1.68 2.11 2.18 2.33 2.45
0.17 0.15 0.27 0.27 0.34
0.94 0.92 0.86 0.87 0.83
1.79 2.46 2.57 2.69 2.67
0.12 0.27 0.15 0.18 0.08
0.92 0.94 0.94 0.94 0.98
2.24 3.37 3.98 3.48 3.3
0.1 0.28 0.18 0.07 0.15
1.0 1.0 1.0 1.0 1.0
2.1 2.69 3.06 3.01 3.07 3.62
0.1 0.11 0.1 0.11 0.15 0.2
0.96 0.98 0.98 0.98 0.96 0.95
2.37 2.41 4.3*
0.35 0.38 0.66*
0.85 0.75 0.99*
2.9 3.27 4.37
0.2 0.21 0.1
0.93 0.94 0.99
3.88 3.6 4.76 2.6
0.1 0.11 0.26 0.13
1.0 1.0 1.0 1.0
3.34 3.89 4.97 2.96 4.19
0.18 0.23 0.42 0.17 0.66
0.95 0.93 0.89 0.93 0.64
0.31 0.37 0.85 0.74 0.63 0.66 1.3 0.54 2.01 1.02 0.61 0.70
0.02 0.05 0.07 0.07 0.05 0.03 0.06 0.05 0.1 0.03 0.02 0.17
0.98 0.95 0.98 0.98 0.98 0.99 0.99 0.98 0.99 0.99 0.99 0.89
Average Rs (n = 4) (mL d−1)
± SD
%CV
1.95 2.66 2.95 2.88 2.87
0.26 0.53 0.77 0.49 0.38
13% 20% 26% 17% 13%
3.12 3.29 4.7 2.78
0.64 0.64 0.31 0.26
20% 19% 7% 9%
a = no significant adsorption. N/A = not measured. *The last data time point (29/08/17) in the calibration for this analyte/site was excluded. Perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluorobutane sulfonate (PFBS), perfluorohexane sulfonate (PFHxS), perfluorooctane sulfonate (PFOS), 6:2 fluorotelomer sulfonate (6:2 FTS), 8:2 fluorotelomer sulfonate (8:2 FTS).
3.2. Uptake of PFASs in PE passive samplers
yields an estimated mean t1/2 for PFASs in groundwater using the PE sampler of 240 days (with a range of 122–490 days for PFOS and PFOA, respectively, noting that modelled estimates using available data were only possible for PFHpA, PFHxA, PFOS, PFBS, PFOS, Table S5). Since the time required for the sampler to reach equilibrium according to Eq. (1) is strictly infinite, we define an effective equilibrium time (teq) as the finite time for Cs to reach 99% of its equilibrium value, i.e. teq = 4.605msKsw/Rs. Using the above data for ms, Ksw and Rs we estimate 1611 days (or 4.4 years) to achieve equilibrium with 810 and 3270 days for PFOS and PFOA, respectively. Hence, in comparison with the POCIS that reported t1/2 of up to only 13 days and teq of up to 60 days, the linear range for the sampling of PFASs was significantly extended. The configuration and composition of the PE sampler was successful at slowing down the uptake of PFASs. The increase in the linear range of sampling means that samplers can be deployed to accumulate PFASs for extended periods that can be extremely beneficial given the persistence and stability of these analytes in water systems. The calculated LODs for PFASs with the PE sampler ranged from 0.003 – 0.001 ng mL−1 (based on sampling rates of 2-5 mL d−1 over 15 days and an instrumental average LOD of 0.1 μg mL−1) representing adequate sensitivity.
Uptake of PFASs in PE passive samplers was observed for the five sites investigated (i.e. Sites 1–5, Fig. 1). A different uptake trend was observed for Site 5, therefore the discussion immediately following is in reference to the results observed from Sites 1–4, while Site 5 is discussed separately and in more detail, including an examination of limitations, further on. Linear accumulation of twelve PFASs (r2 ≥ 0.84 for most PFASs; Table 1) in the PE passive samplers was observed over the 82 or 83 day deployment periods (Eq. (2) and Fig. 1). The masses of PFASs accumulated in PEs ranged from 0.0003 μg sampler−1 for PFDA (at day 14) to 1.6 μg sampler−1 for PFHxS (at day 82). As indicated by the grab sampling data, mass of PFASs accumulated was highest at Sites 4 and 5, followed by Site 3 > Site 1 > and Site 2. The predominant PFASs detected in samplers across the sites were PFOS and PFHxS followed by PFOA, consistent with the grab sampling data. In some cases, linear accumulation of PFASs (i.e. PFNA, PFDA, 6:2 FTS and 8:2 FTS) (Fig. 1) was observed in PE samplers where there was no detection in the grab samples due to the relatively low levels of these analytes in groundwater. PFNA linearly accumulated in PE samplers at four sites: Sites 1–4 (Fig. 1), but was only reportable above LOD in grab samples at two sites: Sites 4 and 5 (sites with the highest ΣPFAS concentration). Similarly, PFDA was below LOD in all grab water samples but was linearly accumulated in PE samplers at Sites 4 and 5. This highlights the benefit of the PE passive samplers at the lower range of detection limits where enrichment from water over a two-week period was sufficient to reach reporting capabilities.
3.4. Sampling rates of PFASs in PE passive samplers Sampling rates (Rs) for PFASs at Sites 1–4 were estimated from regression analysis of the linear uptake data (Eq. (2)) and ranged from 1.68 to 4.97 mL d−1 for PFBA and PFOS, respectively (Table 1). Sampling rate data for individual PFASs from Sites 1 to 4 exhibited little variance (%CV = < 26%, Table 1). Consequently, relative sampling rates were consistent across these sites. For example, Rs for PFOA 1.6 ± 0.12 times lower than for PFOS, which consistently had the highest sampling rate (4.7 ± 0.3 mL d−1). As stated previously, despite accumulation of some PFASs (e.g. PFNA, PFDA, 6:2 FTS and 8:2 FTS) in the passive sampler, Rs could not be calculated as levels were below the LOD in grab samples. The mean Rs for PFAS in this study (approximately 3 mL d−1) is some 90 times lower than that estimated for the modified POCIS passive sampler (approximately 270 mL d−1) despite a relatively small difference in the sampler’s surface area (10 cm2 and 16 cm2, respectively) [15]. A reduced Rs is desirable for a PFAS sampler in groundwater
3.3. Estimation of the duration of the linear uptake mode of the PE sampler An important factor for the applicability of the sampler is its effective deployment duration. For a sampler in linear uptake or kinetic mode, this is the half-time (t1/2) i.e. the time it takes for Cs (PFASs) to be half its equilibrium value of KswCw. Linear uptake mode of PFASs is desirable as it is only possible to calculate time-weighted-average water concentration estimates from passive samplers within this sampling phase window. To derive t1/2 (Eq. (3)) the mass of sorbent (Strata XAW) in samplers is 4 × 10−4 kg, mean Ksw 2642 L kg-1 [15]) and the mean sampling rate (Rs) as determined in this study 0.003 L d-1. This 427
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
sorption coefficient and ks the PFASs mass transfer coefficient. Alvarez et al. adopt a similar approach for POCIS in describing trans-membrane flux of diffusants as a weighted average of transport through both the polymer matrix and aqueous pores [27]. With the POCIS, a decrease in Rs of a factor of 1.5 with molar mass for the same suite of PFASs as employed in the current work was observed. This was taken as evidence that uptake was WBL controlled. However the present direct linear relationship (Fig. 2) implies a more complex situation that may involve the PE as well as aqueous barriers to mass transfer. The relatively small Rs values determined with the PE sampler are of great advantage in groundwater applications, as this means minimal analyte depletion in the system. In most groundwater situations, where significant flow or turbulence is unlikely, depletion of the water in the vicinity of the sampler due to essentially stagnant conditions may become a problem. Although the reproducibility of Rs data between sampling sites in this study is promising and may suggest the influence of reduced flow rates, we were not able to deploy monitors in the narrow groundwater bores to characterise local flow rates. Further investigations under a range of different flow conditions would further validate the diffusion characteristics and flow effects on PFASs sampling rates when using this PE passive sampler.
Fig. 2. Plot of sampling rate (Rs, mL d−1) versus molar mass (g mol−1) for those PFASs where parallel grab sampling data was available.
environments as it contributes to a longer linear uptake phase and extended monitoring capabilities. With the present PE sampler, a direct linear relationship (r2 = 0.92) of Rs with PFASs molar mass was evident, such that an increase in Rs by a factor 1.6 was observed over the PFASs molar mass range of 314–514 g mol−1 (Fig. 2). As well as facilitating the estimation of sampling rates for uncalibrated PFASs, this relationship may provide insight into the mass transfer process operating under the conditions at the groundwater sites. If the accumulation of PFASs can be considered as a multi-stage process through a water boundary layer (WBL) and PE matrix or aqueous pores in the PE to the sorbent, the overall mass transfer coefficient (ko = Rs/A, where A is the surface area of the sampler) can be represented by [15,24]
1 A = = w + ko Rs Dw
p
Dw + (1
2
) Dp Kpw
2
+
1 Ksw ks
3.5. Atypical sampling behaviour and recognition of limitations The PE samplers at Site 5 exhibited substantially lower sampling rates (by a factor of approximately 4) when compared to all other sites (Table 1). While concentrations of the PFASs determined from the grab samples appear to be more variable at this site, this variability cannot readily explain the reduced sampling rates observed. It is noted that uptake rates appear linear during the first four sampling periods with a noticeable depletion of the sampler towards the last sampling point (i.e. 29/08/17). The coefficient of variation for derived water concentrations between triplicate PE samplers at this last sampling time point were considerably higher (%CV = 39%) in comparison with the average of all other PE sampler triplicates (%CV = 14%). The characteristics of this groundwater bore were notably different, being characterised by distinct hydrocarbon odour (likely due to its central location and adjacent activities). It was also noted during the penultimate sampling period that some training activity was conducted adjacent to the borehole that used an unknown quantity of fresh water on the surface. Currently, we have no suitable explanation for the atypical sampling behaviour at this site. It is possible that the high concentration of hydrocarbons at Site 5 has influenced uptake, however the
(4)
where the first term represents resistance to mass transfer in the WBL (δw is WBL thickness and Dw the PFASs diffusion coefficient in water). The second term describes parallel resistance to PFASs transfer in PE and aqueous pores with δp (2 mm) being PE thickness, Dp the PFASs diffusion coefficient in PE, Kpw the PE/water distribution coefficient of the PFASs and θ and the PE tortuosity and porosity respectively. The final term represents the sorbent phase. In this, Ksw is the PFASs
Fig. 3. Mean concentrations of PFASs in groundwater from PE samplers (ng L−1; n = 3) vs Grab samples (n = 1) at all sites, for all PFASs. The dotted red line represents a 1:1 ratio (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article).
428
429
Grab Sample
37 38 48 38 48 18 5.7 8.4 11 179 39 997 7.6 5.4 6.2 < LOD 1.1
Ave
0.4
3.9 6.4 9.4 4.9 5.0 1.7 0.3 3.2 4.1 52 16 173 2.1 0.9 1.2
± SD
117 150 181 139 270 73 19 31 40 660 66 4865 24 40 20 < LOD < LOD
38 47 62 53 71 28 6.6 14 21 337 67 848 2.3 3.3 1.2 < LOD 0.3
Ave
0.1
1.4 1.2 1.6 5.7 7.3 4.9 0.8 0.5 3.1 95 7.4 153 0.7 0.8 0.2
± SD
PE Samplers (n = 3)
118 223 1084 124 88 0.1 < LOD 684 775 2619 8 N/A < LOD < LOD 0.3 < LOD < LOD
PE Samplers (n = 3)
< LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD
Ave
Site E
0.16 0.01 0.09
0.29
± SD
Site D
0.39 < LOD 0.17 0.06 0.24 < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD < LOD
Ave
72 105 142 89 168 27 < LOD 41 44 703 56 3816 < LOD < LOD < LOD < LOD < LOD
Grab Sample
0.1
6 17 70 8 9 0.03 < LOD 31 22 161 1.5 3
± SD
PE Samplers (n = 3)
PE Samplers (n = 4)
Grab Sample
Site A
Blanks
1241 2894 11270 3099 6203 96 2.0 4208 4808 36399 2668 61320 26 1.8 317 2.3 1.2
Ave
359 763 347 384 526 117 < LOD 73 63 315 36 319 < LOD < LOD 1.2 < LOD < LOD
Ave
0.2
15 23 12 13 90 12 < LOD 4.0 7.6 50 3.4 5.1
± SD
357 532 259 271 464 77 < LOD 47 38 302 25 550 < LOD < LOD < LOD < LOD < LOD
Grab Sample
159 222 876 286 684 26 0.3 295 511 5816 314 9311 3.9 0.3 28 0.1 0.1
± SD
PE Samplers (n = 3)
PE Samplers (n = 3)
Site F
119 191 1207 120 180 < LOD < LOD 926 682 2711 14 1160 < LOD < LOD < LOD < LOD < LOD
Grab Sample
Site B
154 272 102 199 272 104 < LOD 12 9.1 78 35 286 < LOD < LOD < LOD < LOD < LOD
Ave
14 31 8.4 29 18 7.2 < LOD 1.8 1.1 5.0 4.3 19
± SD
PE Samplers (n = 3)
Site G
2824 3768 13057 2499 5433 58 < LOD 5091 6186 34178 4995 49144 < LOD 30 325 < LOD < LOD
Grab Sample
199 250 85 124 262 64 < LOD < LOD < LOD 77 21 561 < LOD < LOD < LOD < LOD < LOD
Grab Sample
80 129 181 130 154 69 23 40 64 700 127 4015 24 29 40 < LOD 0.7
Ave
0.04 0.04 0.03 0.01 0.04 0.03 0.02 0.03 0.03 0.04 0.02 0.03 0.02 0.06 0.06 0.23 0.05
LOD (ng mL−1)
0.8
2.9 9.4 23 19 37 12 3.5 6.8 11 101 0.7 869 23 6.9 6.9
± SD
PE Samplers (n = 3)
Site C
0.14 0.13 0.11 0.04 0.12 0.09 0.06 0.11 0.10 0.13 0.07 0.10 0.07 0.18 0.19 0.77 0.17
LOQ (ng mL−1)
123 153 211 134 293 67 15 49 55 856 99 6240 < LOD 69 29 < LOD < LOD
Grab Sample
< LOD - below limit of detection. N/A - could not be quantified due to chromatographic interference. perfluorobutanoic acid (PFBA), perfluoropentanoic acid (PFPeA), perfluorohexanoic acid (PFHxA), perfluoroheptanoic acid (PFHpA), perfluorooctanoic acid (PFOA), perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluorobutane sulfonate (PFBS), perfluoropentane sulphonate (PFPeS), perfluorohexane sulfonate (PFHxS), perfluoroheptane sulphonate (PFHpS), perfluorooctane sulfonate (PFOS), perfluorononane sulfonate (PFNS), 8:2 fluorotelomer sulfonate (8:2 FTS), 6:2 fluorotelomer sulfonate (6:2 FTS), 4:2 fluorotelomer sulfonate (4:2 FTS), perfluorooctane sulfonamide (FOSA).
PFBA PFPeA PFHxA PFHpA PFOA PFNA PFDA PFBS PFPeS PFHxS PFHpS PFOS PFNS 8:2 FTS 6:2 FTS 4:2 FTS FOSA
PFASs (ng L−1)
PFBA PFPeA PFHxA PFHpA PFOA PFNA PFDA PFBS PFPeS PFHxS PFHpS PFOS PFNS 8:2 FTS 6:2 FTS 4:2 FTS FOSA
PFASs (ng L−1)
Table 2 Comparison of water concentration estimates of 17 PFASs from PE passive samplers and grab samples (including limits of detection (LODs) and quantification (LOQ)), from a validation study conducted adjacent to a fire training ground, between 27.03.2018 - 28.6.2018.
S.L. Kaserzon et al.
Journal of Hazardous Materials 366 (2019) 423–431
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al.
mechanisms that influence this interaction are currently poorly understood [28].
Acknowledgements QAEHS is a joint venture of The University of Queensland and Queensland Health and Forensic Scientific Services (QHFSS). We would like to acknowledge Ben Tscharke, Robert Gray, Kristie Thompson, Gabriele Elisei, Sean Van Niekerk and Jake O’Brien for assistance with sampler deployment and retrieval. Part of this study was supported by Australian Research Council (ARC) Linkage grant LP160100510.
3.6. Validation trial of PE samplers and comparison with grab samples A field deployment was conducted in order to validate PE sampler performance by using the previously determined calibration data for a different groundwater contamination scenario and comparing samplerderived concentrations with those from grab sampling. PE samplers were deployed at 7 groundwater sites that receive various inputs from AFFF activities (Table S2). Good correlation was observed when plotting data from all sites and PFASs investigated (r2 = 0.98, Fig. 3), even across the wide range of contamination levels observed between the sites. For example PFOS and PFHxS concentrations ranged from 0.3 and 0.1 ng mL−1 (Site G) to 60 and 36 ng mL−1 (Site B), respectively (Table 2). Fifteen PFASs were detected in grab samples, while 17 were detected in PE passive samplers. In particular, the PE sampler was more effective at detecting precursor PFASs (i.e. 4:2 FTS, and FOSA present at relatively low levels (< 2.3 ng L−1). Mean sampling rates from the PE sampler calibration study (i.e. Table 1) were used to estimate PFASs water concentrations at the field deployment site (based on Eq. (2)). Where sampling rate data was not available (i.e. for PFNA, PFDA, PFPeS, PFHpS, PFNS, 8:2 FTS, 4:2 FTS and FOSA), modelled sampling rates were used based on the sampling rate (Rs, L d−1) versus molar mass (g mol−1) relationship depicted in Fig. 2. Focussing on data from individual sites, a linear correlation between PE samplers and grab sample derived concentrations at individual sites of with r2 values of > 0.96 was observed for five sites, and 0.68 and 0.76 observed for Sites G and F respectively. This decreased agreement in the latter two cases case may be the result of the close vicinity of these two adjacent wells to a marine embayment and consequent tidal influence. Depending on when a single point-in-time grab sample is collected within the tidal cycle may cause variability. It should be remembered that the PE samplers provide a time-weighted average water concentration estimate over 93 days, while the grab samples represent a single point in time viz. concentrations at sampler retrieval. Despite this, overall, a close correlation between PFASs concentrations with the two methods was apparent.
Appendix A. Supplementary data Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.jhazmat.2018.12.010. References [1] Z. Wang, J.C. Dewitt, C.P. Higgins, I.T. Cousins, A Never-Ending Story of Per- and Polyfluoroalkyl Substances (PFASs)? Environ. Sci. Technol. 51 (2017) 2508–2518. [2] R.C. Buck, P.M. Murphy, M. Pabon, Chemistry, properties, and uses of commercial fluorinated surfactants, in: P.T. Knepper, T.F. Lange (Eds.), Polyfluorinated Chemicals and Transformation Products, Springer, Berlin Heidelberg, Berlin, Heidelberg, 2012, pp. 1–24. [3] J. Bräunig, C. Baduel, A. Heffernan, A. Rotander, E. Donaldson, J.F. Mueller, Fate and redistribution of perfluoroalkyl acids through AFFF-impacted groundwater, Sci. Total Environ. 596-597 (2017) 360–368. [4] E.F. Houtz, C.P. Higgins, J.A. Field, D.L. Sedlak, Persistence of perfluoroalkyl acid precursors in AFFF-Impacted groundwater and soil, Environ. Sci. Technol. 47 (2013) 8187–8195. [5] A. Kärrman, K. Elgh-Dalgren, C. Lafossas, T. Møskeland, Environmental levels and distribution of structural isomers of perfluoroalkyl acids after aqueous fire-fighting foam (AFFF) contamination, Environ. Chem. 8 (2011) 372–380. [6] L. Ahrens, Polyfluoroalkyl compounds in the aquatic environment: a review of their occurrence and fate, J. Environ. Monit. 13 (2011) 20–31. [7] I.T.C.K. Prevedouros, R.C. Buck, S.H. Korzeniowski, Sources, fate and transport of Perfluorocarboxylates, Environ. Sci. Technol. 40 (2006). [8] T.M.Jennifer L. Guelfo, David M. Klein, David A. Savitz, Scott Frickel, Michelle Crimi, Eric M. Suuberg, Evaluation and management strategies for perand polyfluoroalkyl substances (PFASs) in drinking water aquifers: perspectives from impacted U.S. Northeast communities, Environ. Health Perspect. 126 (2018). [9] C. Ort, M.G. Lawrence, J. Rieckermann, A. Joss, Sampling for pharmaceuticals and personal care products (PPCPs) and illicit drugs in wastewater systems: are your conclusions valid? A critical review, Environ. Sci. Technol. 44 (2010) 6024–6035. [10] C. Baduel, C.J. Paxman, J.F. Mueller, Perfluoroalkyl substances in a firefighting training ground (FTG), distribution and potential future release, J. Hazard. Mater. 296 (2015) 56-53. [11] D. Szabo, T.L. Coggan, T.C. Robson, M. Currell, B.O. Clarke, Investigating recycled water use as a diffuse source of per- and polyfluoroalkyl substances (PFASs) to groundwater in Melbourne, Australia, Sci. Total Environ. 644 (2018) 1409–1417. [12] ISO Water quality, Sampling Part 23: Guidance on Passive Sampling in Surface Waters, ISO 5667-23:2011, ISO, Genève, Switzerland, 2016 in. [13] C. Harman, I.J. Allan, E.L.M. Vermeirssen, Calibration and use of the polar organic chemical integrative sampler-a critical review, Environ. Toxicol. Chem. 31 (2012) 2724–2738. [14] B. Vrana, I.J. Allan, R. Greenwood, G.A. Mills, E. Dominiak, K. Svensson, J. Knutsson, G. Morrison, Passive sampling techniques for monitoring pollutants in water, TrAC Trends Anal. Chem. 24 (2005) 845–868. [15] S.L. Kaserzon, K. Kennedy, D.W. Hawker, J. Thompson, S. Carter, A.C. Roach, K. Booij, J.F. Mueller, Development and calibration of a passive sampler for perfluorinated alkyl carboxylates and sulfonates in water, Environ. Sci. Technol. 46 (2012) 4985–4993. [16] S.L. Kaserzon, E.L. Vermeirssen, D.W. Hawker, K. Kennedy, C. Bentley, J. Thompson, K. Booij, J.F. Mueller, Passive sampling of perfluorinated chemicals in water: flow rate effects on chemical uptake, Environ. Pollut. 177 (2013) 58–63. [17] S.L. Kaserzon, D.W. Hawker, K. Booij, D.S. O’Brien, K. Kennedy, E.L. Vermeirssen, J.F. Mueller, Passive sampling of perfluorinated chemicals in water: in-situ calibration, Environ. Pollut. 186 (2014) 98–103. [18] K. Booij, N.L. Maarsen, M. Theeuwen, R. Van Bommel, Method to account for the effect of hydrodynamics on polar organic compound uptake by passive samplers, Environ. Toxicol. Chem. 36 (2017) 1517–1524. [19] J.K. Challis, M.L. Hanson, C.S. Wong, Development and calibration of an organicdiffusive gradients in thin films aquatic passive sampler for a diverse suite of polar organic contaminants, Anal. Chem. 88 (2016) 10583–10591. [20] D. Guan, Y. Li, N. Yu, G. Yu, S. Wei, H. Zhang, W. Davison, X. Cui, L. Ma, J. Luo, In situ measurement of perfluoroalkyl substances in aquatic systems using diffusive gradients in thin-films technique, Water Res. 144 (2018). [21] S. Bopp, H. Weiß, K. Schirmer, Time-integrated monitoring of polycyclic aromatic hydrocarbons (PAHs) in groundwater using the Ceramic Dosimeter passive sampling device, J. Chromatogr. A 1072 (2005) 137–147. [22] J. Cristale, A. Katsoyiannis, C. Chen, K.C. Jones, S. Lacorte, Assessment of flame retardants in river water using a ceramic dosimeter passive sampler, Environ. Pollut. 172 (2013) 163–169.
4. Conclusion The investigated microporous PE diffusion passive sampler represent the first passive sampling solution suitable for quantitative long-term monitoring of PFASs in groundwater. In calibration studies, twelve PFASs were detected in-situ in PE samplers, with linear accumulation observed over a three-month deployment period. Sampling rates (Rs) determined for PFASs (mean of approximately 3 mL d−1) were consistent at four of the five calibration sites. A field validation conducted at seven contaminated groundwater sites, using the previously determined calibration parameters, showed successful quantification of PFASs over a wide range of concentrations, and with generally good agreement with grab sampling data. Successful quantification of precursor PFASs (i.e. 4:2 FTS and FOSA) was also possible with the PE samplers, despite concentrations in grab samples being < LOD. The sampler’s relatively simple, robust and compact design facilitates deployment in narrow groundwater bores or other confined systems. In comparison with previous types of samplers, the duration of the linear uptake phase of PFASs in this design was successfully extended, as evidenced by a mean t1/2 estimate of 240 days. This makes the sampler useful for longer-term deployments (3 months or more) and for temporal as well as spatial assessments of PFASs groundwater concentrations. However, further investigation of the sampler’s performance in longer deployments of 6–12 months duration, under different physicochemical conditions and in additional environmental matrices, such as surface and wastewaters, would be beneficial to establish its wider applicability. 430
Journal of Hazardous Materials 366 (2019) 423–431
S.L. Kaserzon et al. [23] H. Franquet-Griell, V. Pueyo, J. Silva, V.M. Orera, S. Lacorte, Development of a macroporous ceramic passive sampler for the monitoring of cytostatic drugs in water, Chemosphere 182 (2017) 681–690. [24] V. Fauvelle, N. Montero, J.F. Mueller, A. Banks, N. Mazzella, S.L. Kaserzon, Glyphosate and AMPA passive sampling in freshwater using a microporous polyethylene diffusion sampler, Chemosphere 188 (2017) 241–248. [25] J.N. Huckins, J.D. Petty, K. Booij, Monitors of Organic Chemicals in the Environment: Semipermeable Membrane Devices, Springer Science & Business Media, 2006. [26] C. Baduel, J.F. Mueller, A. Rotander, J. Corfield, M.J. Gomez-Ramos, Discovery of novel per- and polyfluoroalkyl substances (PFASs) at a fire fighting training ground
and preliminary investigation of their fate and mobility, Chemosphere 185 (2017) 1030–1038. [27] D.A. Alvarez, J.N. Huckins, J.D. Petty, T.L. Jones-Lepp, F. Stuer-Lauridsen, D.T. Getting, J.P. Goddard, A. Gravell, Tool for monitoring hydrophilic contaminants in water: polar organic chemical integrative sampler (POCIS), in: R. Greenwood, G. Mills, B. Vrana (Eds.), Passive Sampling Techniques in Environmental Monitoring, Comprehensive Analytical Chemistry, 48, Elsevier, Amsterdam, 2007, pp. 171–197. [28] J.L. Guelfo, C.P. Higgins, Subsurface transport potential of perfluoroalkyl acids at aqueous film-forming foam (AFFF)-impacted sites, Environ. Sci. Technol. 47 (2013) 4164–4171.
431