Fate of Perfluorinated Carboxylates and Sulfonates During Snowmelt

ARTICLE pubs.acs.org/est

Fate of Perfluorinated Carboxylates and Sulfonates During Snowmelt Within an Urban Watershed Torsten Meyer,*,† Amila O. De Silva,‡ Christine Spencer,‡ and Frank Wania† †

Department of Physical and Enironmental Sciences, University of Toronto Scarborough, 1265 Military Trail, Toronto, Ontario, Canada M1C 1A4 ‡ Water Science and Technology Directorate, Environment Canada, 867 Lakeshore Road, P.O. Box 5050 Burlington, Ontario, Canada L7R 4A6

bS Supporting Information ABSTRACT: The transport dynamics of perfluorinated carboxylic acids and sulfonates during snowmelt in the highly urbanized Highland Creek watershed in Toronto, Canada was investigated by analyzing river water, bulk snow, and groundwater, sampled in February and March 2010, by means of liquid chromatography-tandem mass spectrometry. Perfluorohexanoate, perfluorooctanoate, and perfluorooctane sulfonate were dominant in river water, with concentrations of 4.0
14 ng 3 L
1, 2.2
7.9 ng 3 L
1, and 2.1
6.5 ng 3 L
1, respectively. Relatively high levels of perfluorohexanoate may be related to the recent partial replacement in various consumer products of perfluorooctyl substances with shorter-chained perfluorinated compounds (PFCs). Highest PFC concentrations were found within the more urbanized part of the drainage area, suggestive of residential, industrial, and/or traffic-related sources. The riverine flux of PFCs increased during the snowmelt period, but only approximately one-fifth of the increased flux can be attributed to PFCs present in the snowpack, mostly because concentration in snow are generally quite low compared to those in river water. The remainder of the increased flux must be due to the mobilization of PFCs by the high flow conditions prevalent during snowmelt. Run-off behavior was clearly dependent on perfluoroalkyl chain length: Dilution with relatively clean snowmelt water caused a drop in the river water concentrations of short-chain PFCs at high flow during early melting. This prevented an early concentration peak of those water-soluble PFCs within the stream, as could have been expected in response to their early release from a melting snowpack. Instead, concentrations of particle-associated long-chain PFCs in creek water peaked early in the melt, presumably because high flow mobilized contaminated particles from impervious surfaces in the more urbanized areas of the watershed. The ability to enter the subsurface and deeper groundwater aquifers increased with the PFCs0 water solubility, that is, was inversely related to perfluoroalkyl chain length.

’ INTRODUCTION Perfluorinated carboxylic acids (PFCAs) and sulfonates (PFSAs) are persistent, anthropogenic contaminants with ubiquitous environmental distribution.1 Important point sources of PFCs are wastewater treatment plants,2 fluorochemical manufacturing facilities, places of fire fighting foam usage, and landfills.3 Notable nonpoint sources are urban areas,4 as well as atmospheric deposition following oxidation of volatile precursors.5 Nonionic precursors such as perfluoroalkyl sulfonamides, perfluoroalkyl sulfonamido ethanols, and fluorotelomer alcohols are relatively volatile and can undergo atmospheric transport prior to transformation and deposition.5 In urban regions however, atmospheric transformation of precursors to PFCs is limited due to the presence of relatively high NOx levels.6,7 Significant sinks and reservoirs of PFCs are sediments and deep ocean waters.3,8 Whereas shorter-chained PFCs are relatively water-soluble, those with longer chains can sorb strongly to particles and are subject to sedimentation and resuspension processes.8 r 2011 American Chemical Society

Whereas we are not aware of any study having investigated the fate of PFCs in urban watersheds during snowmelt, there are reports of concentrations of PFCs during rain events within urban streams,9,10 lakes,7 and street runoff,11 as well as PFCs in urban snow.12,13 Most of those studies related the presence of PFCs largely to urban activities, and to a lesser extent to atmospheric sources. Murakami and Takada4 found a correlation between PFCs in major rivers in Japan and population density. Besides point sources, urban watersheds comprise significant nonpoint sources of PFCs to surface waters.7,9,10 The amount of PFCs released from urban nonpoint sources to Japanese streams was estimated to be similar to the amount released from Special Issue: Perfluoroalkyl Acid Received: January 11, 2011 Accepted: June 2, 2011 Revised: May 26, 2011 Published: June 21, 2011 8113

dx.doi.org/10.1021/es200106q | Environ. Sci. Technol. 2011, 45, 8113–8119

Environmental Science & Technology wastewater treatment plants.14 Notably higher concentrations of PFCs in urban streams during daytime were attributed to enhanced human outdoor activity.10 Other potential sources are parking lots,7 fine street particles originating from heavily used streets,4 commercial and transport related land use,14 as well as indoor air from residential and commercial buildings.15 During peak runoff the PFC fluxes in the streams notably increase,9,10 indicative of their fast removal from watershed surfaces. In urban surface water short- to medium-chained PFCs such as perfluorooctanoate (PFOA),7,16 perfluorooctane sulfonate (PFOS),9,16 and perfluorononanoate (PFNA)10 are dominant. Similarly, relatively high concentrations in urban snow (Dalian and Shenyang, China) have been reported for PFOS and PFOA;12,13 however, also for PFHpA.13 The fate of organic chemicals in urban watersheds is highly complex, especially when snow and snowmelt processes are involved.17 Many organic chemicals can be efficiently scavenged from the atmosphere by snow and may be released during melting, many weeks after deposition, in the form of short and concentrated pulses (e.g., refs 18 and 19). Differences between stormwater runoff and snowmelt water ablation include the timing of chemical transport and the amplification of concentrations. In urbanized areas snow melts earlier and meltwater arrives faster at the streams than in green spaces.20 Also, organic chemicals, including PFCs, are released from melting snow packs at different times: the water-soluble fraction often elutes early during the melt period, whereas the particulate fraction is released at the end of melting.21,22 During runoff in urban areas chemicals are rapidly removed from impervious surfaces along channels and canalized stream beds. When runoff occurs on natural ground, contaminants can be transported with overland or subsurface flow, whereby the former delivers contaminants to the streams relatively fast and undiluted. This study aimed at identifying the environmental pathways of PFCs during snowmelt runoff within a highly urbanized watershed by integrating data of chemical concentrations in bulk snow, river water and groundwater with hydrological data such as river flow rates and precipitation. We previously investigated the transport dynamics of polycyclic aromatic hydrocarbons and pesticides during snowmelt in the same watershed.17 The two studies complement each other by providing insight into the fate of a variety of organic contaminants under different snowmelt conditions.

’ MATERIALS AND METHODS Site Description. Highland Creek flows through the eastern part of the city of Toronto and discharges into Lake Ontario. Its watershed (102 km2) is highly urbanized (ca. 85% is used residentially, industrially/commercially, and for roads 23), with the remainder being green space (Figure 1). The only wastewater treatment plant within the watershed (Highland Creek WWTP) has its outfall submerged within Lake Ontario. None of the sampling sites receives water that is influenced by known point sources of PFCs. The area southeast of the former shoreline of ancient Lake Iroquois consists largely of a sand plain which allows for extensive exchange between surface and subsurface water,23 suggesting the potential of transfer of watersoluble pollutants to groundwater aquifers. The western part of the watershed is upstream of the eastern part. The former is more heavily urbanized than the latter, that is, it comprises more

ARTICLE

Figure 1. Highland Creek watershed with sampling locations and the gauging stations GS-E (East) and GS-W (West) (green color on the map designates green space). Sites A to E, river water samples; site GW, groundwater samples.

densely populated areas, and land use related to industry, and utilities/transportation. Sampling. River water, bulk snow and groundwater were sampled during one snowmelt period in February/March 2010, lasting nearly two weeks. At the end of this period most of the snow around the watershed had melted. Additional river water samples, representing normal flow conditions, were taken once after the melt period (Figure 2). The period of snowpack accumulation prior to the melt was relatively short and not interrupted by melt-freeze cycles (Figure 2 and Supporting Information (SI) Figure S1). River water was sampled 13 times at site A, and 4 times at four additional sites (sites B
E) (Figures 1 and 2). Samples were collected in one-liter high density polyethylene (HDPE) bottles according to the procedure described in USGS.24 Simultaneously collected, smaller samples were analyzed for specific conductivity (SI Figures S2
S4). Bulk snow samples were collected in four-liter HDPE bottles on three different days at site A and once at sites B and C by using a solvent-rinsed stainless steel shovel to excavate quader-formed bulk snow blocks comprising the entire snowpack depth. Groundwater samples (GW) were collected twice in the middle of the melt period, and once at the end of melt (Figure 2) at the Scarborough Bluffs, a cliffed shoreline of Lake Ontario located southeast of the watershed, where groundwater from two horizontal aquifers seeps out of the cliff slope (Figure 1 and SI Figure S5).25 Three and two samples were taken from the lower and upper aquifer, respectively, by digging trenches into the cliff slope. GW flow in the Highland Creek watershed is mainly from North to South.26 The depth of the water table near the Bluffs is usually 1
2 m. At the sampling site the aquitard separating both aquifers ranges between approximately 4 and 10 m depth, whereas the lower aquifer reaches down to 30 m.25 The sampling trenches at the cliff slope were located near the bottom of each of the aquifers. Products containing fluorinated polymers were avoided during sampling. All samples were stored at 4 °C until extraction. Meteorological records on an hourly basis provided by the weather station of the University of Toronto Scarborough (near 8114

dx.doi.org/10.1021/es200106q |Environ. Sci. Technol. 2011, 45, 8113–8119

Environmental Science & Technology

ARTICLE

ammonia in methanol under gravity and by using a vacuum pump at the very end. Whereas the first fraction elutes PFOSA, the second fraction elutes all other compounds. The extracts were concentrated to aliquots of 0.25 mL using N2. Perfluorohexanoate (PFHxA), perfluoroheptanoate (PFHpA), PFOA, PFNA, perfluorodecanoate (PFDA), perfluoroundecanoate (PFUnA), perfluorododecanoate (PFDoA), perfluorotridecanoate (PFTrA), perfluorotetradecanoate (PFTeA), perfluorobutanesulfonate (PFBS), perfluorohexanesulfonate (PFHxS), PFOS, and perfluorooctanesulfonamide (PFOSA) (SI Table S1) were quantified in the extracts using an Agilent 1100 liquid chromatograph (Agilent Technology, PaloAlto, CA), coupled to an API 4000 triple quadrupole mass spectrometer (Applied Biosystems, Foster City, CA) operated in electro-spray ionization mode (ESI, negative polarity) with multiple reaction monitoring (MRM). Method detection limits (MDLs) are listed in SI Table S2. Quality assurance involved the extraction of two method blanks with every set of samples, and two field blanks. Field blanks were subjected to the same extraction and cleanup procedures as the samples. All reported data were blank corrected using the averages of all method blanks and 2 field blanks, provided they exceeded the MDLs. A detailed QA/QC description is provided within the Supporting Information.

’ RESULTS AND DISCUSSION

Figure 2. River flow rates (m3 3 s
1) at the two gauging stations GS-E and GS-W, temperature (°C), and precipitation events (pictograms) during the melt period in 2010. The sample numbers in the upper two subplots are assigned according to the sampling sites’ nearest gauging station (Figure 1). UA, upper aquifer; LA, lower aquifer; S, bulk snow samples. Samples taken on March 18 represent normal low flow conditions.

site A), stream discharge records on a quarter-hourly basis from two gauging stations operated by Environment Canada27 (Figures 1 and 2), and snowpack depth records from a weather station located 2 km West of the watershed (SI Figure S1) were used to identify the driving forces of the observed phenomena. Analysis. The chemical analytical methods are described in detail elsewhere.28,29 Briefly, duplicate samples of 250 mL each were spiked with isotopically labeled recovery standards and extracted using Oasis WAX solid phase extraction (SPE) cartridge (6 cm3, 150 mg, 30 μm) from Waters (Milford, MA). The first fraction was eluted with 6 mL methanol under gravity only, whereas the second fraction was eluted with 8 mL of 0.1%

Concentrations and Composition in Streamwater. Concentrations in environmental samples are reported in SI Tables S3
S6. The dominant substance in streamwater was PFHxA, followed by PFOA and PFOS (SI Figure S6). The concentration ranges of the PFCs in river water were as follows: 4.0
14 ng 3 L
1 for PFHxA, 1.2
4.5 ng 3 L
1 for PFHpA, 2.2
7.9 ng 3 L
1 for PFOA, 0.80
2.5 ng 3 L
1 for PFNA, 0.33
1.6 ng 3 L
1 for PFDA, 0.070
0.44 ng 3 L
1 for PFUnA, 0.026
0.48 ng 3 L
1 for PFDoA, not detected
0.086 ng 3 L
1 for PFTrA, not detected
0.13 ng 3 L
1 for PFTeA, 0.28
3.8 ng 3 L
1 for PFBS, 0.24
1.7 ng 3 L
1 for PFHxS, 2.1
6.5 ng 3 L
1 for PFOS, and 0.0045
0.32 ng 3 L
1 for PFOSA (Figures 3 and 4). Whereas the predominance of both PFOA and PFOS is similar to that in other urban streams,9,10 the high abundance of PFHxA is unusual. Its concentrations are elevated to a similar extent in eastern and western Highland Creek and in a similar pattern as other shortchained PFCs (Figures 3 and 4) and its prevalence can therefore not be explained by unknown point sources or biotransformation of compounds that are released in association with fire-fighting foam usage. Recent partial replacement of C8
PFCs by shorterchained PFCs for a variety of consumer products may be responsible for its high abundance.30 Relatively high concentrations of PFCs were found at sampling sites C and D, pointing to a relationship with urban land use. Whereas sites A and B receive meltwater from the green space in the eastern part of the watershed, runoff from the highly urbanized area in the western part affects sites C and D, where we previously also found higher concentrations of PAHs.17 Concentrations in Bulk Snow. The concentrations of the PFCs in bulk snow (Figure 5) were lower than in the meltwater receiving streams (Figures 3 and 4). In particular, concentrations of the short-chained PFCs (PFHxA, PFHpA, PFOA, PFBS, PFHxS) were lower by 1 order of magnitude. Consistent with the higher river water concentrations in the more urbanized western Highland Creek, PFC levels in bulk snow from site C were higher than at sites A and B. Repeated sampling at site A 8115

dx.doi.org/10.1021/es200106q |Environ. Sci. Technol. 2011, 45, 8113–8119

Environmental Science & Technology

ARTICLE

Figure 3. River water concentrations at sampling site A vs river flow rates and sample order. Lower right subplot: PFOS and PFBS
left vertical axis, PFHxS and PFOSA—right vertical axis. Numbers on x-axes correspond to the samples A1 to A13 (Figure 2).

(samples S1
S3) indicated an increase of the PFC bulk snow concentrations over the course of the melt period (Figure 5). The concentration increase between the first two sampling days may have been caused by deposition with snow and rain (Figures 2 and 5), whereby the C6
C10 PFCAs and all quantified PFSAs increased the most. Another increase in bulk snow concentrations occurred between the second and third sampling, with the C12
C13 PFCAs showing the strongest increase. Melt water loss and accumulation of particle bound PFCs on top of the snowpack surface, as well as snow sublimation31 may have caused this concentration increase. Surprisingly, concentrations of the shorter-chained PFCs also increased, even though their relatively high water solubility should have led to an early release from the snow packs.22 However, spatial variability of melting in the Highland Creek watershed is high. The bulk snow sampling location at site A, exhibiting low solar exposure, high albedo, and large heat discharge capacity, experienced relatively little meltwater loss. Snow sublimation presumably outweighed the chemical loss that occurred during melting. In order to quantify the contribution of snowmelt water to the PFC burden in Highland Creek, we compared the chemical amount in bulk snow and in river water during melting. First, the overall chemical input flux into the streams during the melt was calculated as the difference between the chemical flux in the river during high and base flow conditions. Those fluxes were calculated by multiplying the river water concentration measured at sampling sites A and C with the river flow rate recorded at the time of sampling at the nearby gauging stations (Figure 1). Second, the maximum chemical flux from the melting snow cover into the river was estimated by multiplying the measured bulk snow concentrations with the river flow that is due to meltwater

runoff. Those calculations were done for five instances during high flow conditions (A3, A5, A6, C1, C2, Figures 1 and 2). On average only one-fifth of all PFCs originates from the snowpack (SI Figure S7). Whereas snow is estimated to contribute less than 10% of the short-chained PFCs (PFHxA, PFHpA, PFBS) in the streamwater, its contribution for longer-chained PFCs such as PFUnA, PFTrA, and PFOS exceeds 30% (SI Figure S7).The remainder must have been due to remobilisation of PFCs induced by the high flow conditions. Phase Partitioning in the Streams. The particulate and dissolved fractions of the samples were not analyzed separately. Instead, correlations between substances were used to categorize their runoff behavior and to draw conclusions regarding phase partitioning. The concentrations of the shorter-chained C6
C9 PFCAs and the PFSAs at site A are strongly and significantly correlated with each other (except for PFBS, mainly because of one outlier). Similarly correlated with each other are the concentrations of the longer-chained C10
C13 PFCAs. PFTeA (C14) was not considered due to a limited number of samples above the MDL. At the same time, correlations between shortand long-chained PFCAs, as well as between the PFSAs and longchained PFCAs are weak or nonexistent, except for the correlation between PFNA and PFDA (C9 and C10) (r = 0.62, p < 0.05). The latter two chemicals obviously constitute a transition in runoff behavior between the C6
C8 PFCAs and the PFSAs on the one hand, and the C11
C13 PFCAs on the other hand. The divergent behavior of the two groups cannot be explained with a different usage of the chemicals in different areas of the watershed. The concentrations of both groups are enhanced to a similar extent at the western sites C and D (Figures 3
5). This transition is therefore assumed to mark the difference between 8116

dx.doi.org/10.1021/es200106q |Environ. Sci. Technol. 2011, 45, 8113–8119

Environmental Science & Technology

Figure 4. River water concentrations at all sampling sites vs river flow rates at site A and sample order. In order to compare site A with the other sites, the columns of site A are represented by the following samples: first column, average of samples A2, A3; second column, average of A4, A5; third column, A9; fourth column, A13 (Figure 2). Column 4 of all sites represents concentrations during normal flow conditions.

Figure 5. Concentrations in bulk snow samples. Lower right subplot: PFOS, left vertical axis; PFHxS and PFOSA, right vertical axis. Samples S1
S3 were taken at site A, S4 at site B, and S5 at site C.

chemicals that are mainly sorbed to suspended particulate matter (C10
C13 PFCAs), and substances that are predominantly present in the dissolved phase of the river water (C6
C9 PFCAs, PFSAs). The correlation between the two chemicals with intermediate partition properties, PFNA and PFDA, suggests that they were present in both the particulate and dissolved phases in notable parts. This transition is also reflected in the pattern of the PFC concentration sequences (Figures 3 and 4). Whereas the

ARTICLE

concentrations of the shorter-chained substances increase during the melt period, those of the longer-chained PFCs rather show a decreasing trend. Higgins and Luthy8 list log KOC values that describe the equilibrium partitioning of PFCs between organic sediment and water. Accordingly, the C6
C9 PFCAs and the PFSAs on the one hand and the C10
C13 PFCAs on the other hand are separated by a threshold log KOC value between 2.57 and 2.76. Neutral organic chemicals exhibiting similar log KOC values are assumed to be almost completely dissolved within the aqueous phase of the streamwater.17 Therefore, in the case of PFCs additional sorption mechanisms besides that of sorption to organic matter may be effective. Zushi and Masunaga10 may have identified a similar threshold for PFC partitioning. Differences in the runoff behavior of PFNA and PFUnA (C9 and C11) were assumed to reflect the chemicals0 different partitioning between suspended particles and the dissolved phase in the streamwater. In Plassmann et al.22 the phase transition between meltwater and snow grains within a melting snowpack, was represented by the C8
C10 PFCAs and the C6
C8 PFSAs. PFCAs and PFSAs with less than 9 and 7 alkylcarbons, respectively, were mostly dissolved in meltwater, whereas longer-chained PFCs were strongly sorbed to the snow. Besides chain length, the partitioning of PFCs between aqueous phase and sorbed phase may also be influenced by certain ions.8 Thus, at the onset of melting when ion concentration in Highland Creek water exhibits very high values (SI Figures S2, S3),17 an additional fraction of PFCs with intermediate partitioning properties may transfer to the particulate phase, and subsequently settle in areas of calmed flow such as near the river mouth. Runoff Dynamics of the PFCs. One motivation for implementing this study was to find out whether concentration peak releases of water-soluble organic chemicals from a melting snowpack21 are reflected in enhanced streamwater concentrations at the beginning of the melt period. The first three samples of the high resolution concentration sequences in Figure 3 suggests a minor first flush of the relatively water-soluble C6
C10 PFCAs and the PFSAs. The relative small fraction of PFCs originating from the snowpack presumably prevented a more pronounced early enrichment of those substances. Also, in comparison to glacial runoff where the meltwater has very limited contact with soil and vegetation,19 runoff in urban, temperate areas is often characterized by dilution and buffering of the meltwater during subsurface flow, as well as a more heterogeneous runoff pattern around the watershed. Water-soluble inorganic ions, however, were highly enriched in the streams at an early stage of melting (SI Figures S2 and S3). Most of them were constituents of deicing salt and were directly and rapidly flushed almost exclusively from the streets and highway into the streams. The chemical flux of all PFCs substantially increased during the early stage of melting, not least because river water flow increased approximately 10-fold (Figure 2 and SI Figures S8 and S9). The transfer of PFCs from watershed surfaces and subsurfaces to the river is controlled by the intensity of melting, which is apparent in a strong and highly significant correlation between river flow rate and chemical flux at site A (SI Table S7). Meyer et al.17 found a similar correlation for PAHs. Whereas the chemical flux of the C11
C14 PFCAs increased by 1 to 2 orders of magnitude, the increase is less than 1 order of magnitude for the shorter-chained C6
C10 PFCAs and the PFSAs (SI Figures S8 and S9). The former are primarily particle-bound and are 8117

dx.doi.org/10.1021/es200106q |Environ. Sci. Technol. 2011, 45, 8113–8119

Environmental Science & Technology more effectively remobilised by high flow. The increase in flux of the shorter-chained PFCs observed here is similar to that observed during rain runoff in streams of Yokohama city, Japan.9,10 The flux increase of the only longer-chained substance (PFUnA) measured by Zushi and Masunaga10 however, is notably smaller than that in our study. The gauging station GS-W receives meltwater from the more urbanized western part of the watershed, whereas GS-E is additionally influenced by runoff from the eastern green space (Figure 1).17 The ratio between the peak river flow rates at GS-E and GS-W provides information about the relative extent of runoff in both parts of the watershed. This ratio was highest on February 27 when the runoff had its peak (SI Figure S10). Melting on that day was mainly induced by a rain event, leading to a relatively uniform runoff pattern around the watershed. On the next day, when melting was induced by irradiation only, GSW recorded relatively higher flow rates (the flow rate ratio GS-E/ GS-W decreased by 13%, reflecting preferred runoff in the urbanized west. At the same time, the concentrations of the long-chained and primarily particle bound PFCs were reaching a peak (Figures 3 and 4). Meyer et al.17 observed a similar correlation between comparably high flow rates at GS-W and elevated PAH levels in Highland Creek streamwater. Accordingly, at an early stage of the melt period, when melting is not induced by rain, the river receives highly polluted runoff primarily from the more urbanized part of the watershed. The peak flow rate ratio increased during subsequent days indicative of the delayed arrival of meltwater from the green space. Linear correlations between river flow rates and chemical fluxes (SI Table S7) and the quarter-hourly flow rate records provided by the gauging station GS-E were used to estimate the total amount of each PFC that passes through sampling site A over the entire melt period (SI Table S8). The ratio between chemical flux and the river flow rate refers to the flow-weighted mean concentration (FWMC). The latter is commonly used to estimate average concentrations of various constituents in rivers. The total amount passing site A ranges between 0.035 g for PFTrA and 13 g for PFHxA. The chemical fluxes at site E located at the river mouth are expected to be similar to those at site A, indicated by similar chemical concentrations at both sites (Figures 1, 3, and 4). Thus, the chemical amounts listed in SI Table S8 should approximate those that had entered Lake Ontario during the investigated snowmelt period. The calculated amounts of PFOA and PFOS are negligible compared to the overall input into Lake Ontario which is dominated by inflow from wastewater treatment plants and Lake Erie.32 PFCs in Groundwater. The concentrations of PFCs in groundwater from the Highland Creek watershed (Figure 6) are well in the range of those found in Tokyo, Japan.33 Transport to groundwater is not limited to water-soluble PFCs, as even relatively hydrophobic substances such as PFDA and PFUnA are readily detected in the lower aquifer (Figure 6). The sand plain prevalent southeast of the former shoreline of Lake Iroquois allows for rapid transfer of meltwater and the chemicals therein from the surface to the upper aquifer (Figure 1).23 At the onset of melting the discharge at the Scarborough Bluff slope may respond within 24 h to the enhanced pressure in the aquifer, imposed by the melt at the surface behind the bluffs.25 Therefore, the PFCs in the samples taken from the upper aquifer several days after the onset of melting likely originated to a large extent at the bluff’s table surface and subsurface. The samples taken from the lower aquifer however, were likely not influenced by this

ARTICLE

Figure 6. Concentrations in groundwater samples taken at the Scarborough Bluffs. Middle subplot: PFNA
left vertical axis, PFDA and PFUnA, right vertical axis. UA-1 and UA-2 taken from the upper aquifer, LA-1 to LA-3 taken from the lower aquifer.

snowmelt period and may originate largely from further inland because the aquitard separating both aquifers near the cliff slope is sparsely permeable. PFC concentration ratios between the streamwater and each of the aquifers were calculated to investigate the relative extent of transfer of individual PFCs to groundwater (SI Figure S11). Although the PFCs in the sampled groundwater were likely not transferred from the streams, the Highland Creek water concentrations may represent typical PFC runoff levels within the watershed. As expected the concentration ratio surface water/lower aquifer is notably higher than that between surface water and upper aquifer, reflecting a more limited chemical transfer to deeper aquifers. Also, both ratios increase with increasing perfluoroalkyl chain length, reflecting the enhanced ability of the more water-soluble PFCs to percolate through soils. Murakami et al.33 also observed a stronger retention of longerchained PFCs within soils. The more hydrophobic PFUnA is the only substance that was equally distributed in both aquifers (Figure 6). This chemical’s presence in the upper aquifer seems less influenced by the snowmelt, and likely represents long-term background concentrations.

’ ASSOCIATED CONTENT

bS

Supporting Information. Additional information, including quality assurance/quality control, specific conductivities, tabulated chemical concentrations, and chemical flux figures.

8118

dx.doi.org/10.1021/es200106q |Environ. Sci. Technol. 2011, 45, 8113–8119

Environmental Science & Technology This material is available free of charge via the Internet at http:// pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT We are indebted to Nick Eyles for his guidance in finding the groundwater sampling site, and Mandy Meriano (both from the University of Toronto Scarborough) for providing insight into the groundwater conditions in the Highland Creek watershed. NSERC is acknowledged for funding. ’ REFERENCES (1) Lau, C.; Anitole, K.; Hodes, C.; Lai, D.; Pfahles-Hutchens, A.; Seed, J. Perfluoroalkyl acids: A review of monitoring and toxicological findings. Toxicol. Sci. 2007, 99, 366–394. (2) Schultz, M. M.; Higgins, C. P.; Huset, C. A.; Luthy, R. G.; Barofsky, D. F.; Field, J. A. Fluorochemical mass flows in a municipal wastewater treatment facility. Environ. Sci. Technol. 2006, 40, 7350–7357. (3) Prevedouros, K.; Cousins, I. T.; Buck, R. C.; Korzeniowski, S. H. Sources, fate and transport of perfluorocarboxylates. Environ. Sci. Technol. 2006, 40, 32–44. (4) Murakami, M.; Takada, H. Perfluorinated surfactants (PFSs) in size-fractionated street dust in Tokyo. Chemosphere 2008, 73, 1172– 1177. (5) Shoeib, M.; Vlahos, P.; Harner, T.; Peters, A.; Graustein, M.; Narayan, J. Survey of polyfluorinated chemicals (PFCs) in the atmosphere over the northeast Atlantic Ocean. Atmos. Environ. 2010, 44, 2887–2893. (6) Ellis, D. A.; Martin, J. W.; De Silva, A. O.; Mabury, S. A.; Hurley, M. D.; Andersen, M. P. S.; Wallington, T. J. Degradation of fluorotelomer alcohols: A likely atmospheric source of perfluorinated carboxylic acids. Environ. Sci. Technol. 2004, 38, 3316–3321. (7) Kim, S.-K.; Kannan, K. Perfluorinated acids in air, rain, snow, surface runoff, and lakes: relative importance of pathways to contamination of urban lakes. Environ. Sci. Technol. 2007, 41, 8328–8334. (8) Higgins, C. P.; Luthy, R. G. Sorption of perfluorinated surfactants on sediments. Environ. Sci. Technol. 2006, 40, 7251–7256. (9) Zushi, Y.; Takeda, T.; Masunaga, S. Existence of nonpoint source of perfluorinated compounds and their loads in the Tsurumi River basin, Japan. Chemosphere 2008, 71, 1566–1573. (10) Zushi, Y.; Masunaga, S. First-flush loads of perfluorinated compounds in stormwater runoff from Hayabuchi River basin, Japan served by separated sewerage system. Chemosphere 2009a, 76, 833–840. (11) Murakami, M.; Shinohara, H.; Takada, H. Evaluation of wastewater and street runoff as sources of perfluorinated surfactants (PFSs). Chemosphere 2009a, 74, 487–493. (12) Liu, W.; Jin, Y.-H.; Quan, X.; Sasaki, K.; Saito, N.; Nakayama, S. F.; Sato, I.; Tsuda, S. Perfluorosulfonates and perfluorocarboxylates in snow and rain in Dalian, China. Environ. Int. 2009a, 35, 737–742. (13) Liu, W.; Dong, G.-H.; Jin, Y.-H.; Sasaki, K.; Saito, N.; Sato, I.; Tsuda, S.; Nakayama, S. F. Occurrence of perfluoroalkyl acids in precipitation from Shenyang, China. Chin. Sci. Bull. 2009b, 54, 2440– 2445. (14) Zushi, Y.; Masunaga, S. Identifying the nonpoint source of perfluorinated compounds: a geographic information system based approach. Environ. Toxicol. Chem. 2009b, 28, 691–700. (15) Gewurtz, S. B.; Bhavsar, S. P.; Crozier, P. W.; Diamond, M. L.; Helm, P. A.; Marvin, C. H.; Reiner, E. J. Perfluoroalkyl contaminants in window film: indoor/outdoor, urban/rural, and winter/summer

ARTICLE

contamination and assessment of carpet as a possible source. Environ. Sci. Technol. 2009, 43, 7317–7323. (16) D’eon, J. C.; Crozier, P. W.; Furdui, V. I.; Reiner, E. J.; Libelo, E. L.; Mabury, S. A. Perfluorinated phosphonic acids in Canadian surface waters and wastewater treatment plant effluent: Discovery of a new class of perfluorinated acids. Environ. Toxicol. Chem. 2009, 28, 2101–2107. (17) Meyer, T.; Lei, Y. D.; Wania, F. Transport of polycyclic aromatic hydrocarbons and pesticides during snowmelt within an urban watershed. Water Res. 2011, 45, 1147–1156. (18) Simmleit, N.; Herrmann, R.; Thomas, W. Chemical behaviour of hydrophobic micropollutants during the melting of snow. IAHS Publ. 1986, 155, 335–346. (19) Lafreni
ere, M. J.; Blais, J. M.; Sharp, M. J.; Schindler, D. W. Organochlorine pesticide and polychlorinated biphenyl concentrations in snow, snowmelt, and runoff at Bow Lake, Alberta. Environ. Sci. Technol. 2006, 40, 4909–4915. (20) Bengtsson, L.; Westerstr€om, G. Urban snowmelt and runoff in northern Sweden. Hydro. Sci. 1992, 37, 263–275. (21) Meyer, T.; Wania, F. Organic contaminant amplification during snowmelt—A review. Water Res. 2008, 42, 1847–1865. (22) Plassmann, M. M.; Meyer, T.; Lei, Y. D.; Wania, F.; McLachlan, M. S.; Berger, U. Laboratory studies on the fate of perfluoroalkyl carboxylates and sulfonates during snow melt. Submitted. (23) TRCA—Toronto and Region Conservation Authority. State of the Watershed Report: Highland Creek Watershed; City of Toronto and the TRCA: Downsview, Ontario, Canada, 1999. (24) USGS—United States Geological Survey. National Field Manual for the Collection of Water-Quality Data, (TWRI—Book 9). Chapter A4—Collection of Water Samples (Version 2.0, 9, 2006). http://pubs. water.usgs.gov/twri9A (accessed Jan. 8, 2010). (25) Eyles, N.; Howard, K. W. F. A hydrochemical study of urban landslides caused by heavy rain: Scarborough Bluffs, Ontario, Canada. Can. Geotech. J. 1988, 25, 455–466. (26) Meriano, M.; Eyles, N. Groundwater flow through Pleistocene glacial deposits in the rapidly urbanizing Rouge River—Highland Creek watershed, City of Scarborough, Southern Ontario, Canada. Hydrogeol. J. 2003, 11, 288–303. (27) Environment Canada—Water Office. Real-time hydrometric data. www.wateroffice.ec.gc.ca (accessed September 16, 2010). (28) Scott, B. F.; Spencer, C.; Lopez, E.; Muir, D. C. G. Perfluoroalkyl acid concentrations in Canadian rivers and creeks. Water Qual. Res. J. Can. 2009, 44, 263–277. (29) Scott, B. F.; De Silva, A. O.; Spencer, C.; Lopez, E.; Backus, S. M.; Muir, D. C. G. Perfluoroalkyl acids in Lake Superior water: trends and sources. J. Great Lakes Res. 2010, 36, 277–284. (30) Ritter, S. K. Fluorochemicals go short. Chem. Eng. News 2010, 88, 12–17. (31) Marsh, P. Snowcover formation and melt: recent advances and future prospects. Hydrol. Proc. 1999, 13, 2117–2134. (32) Boulanger, B.; Peck, A. M.; Schnoor, J. L.; Hornbuckle, K. C. Mass budget of perfluorooctane surfactants in Lake Ontario. Environ. Sci. Technol. 2005, 39, 74–79. (33) Murakami, M.; Kuroda, K.; Sato, N.; Fukushi, T.; Takizawa, S.; Takada, H. Groundwater pollution by perfluorinated surfactants in Tokyo. Environ. Sci. Technol. 2009b, 43, 3480–3486.

8119

dx.doi.org/10.1021/es200106q |Environ. Sci. Technol. 2011, 45, 8113–8119