Environ. Sci. Technol. 2005, 39, 74-79
Mass Budget of Perfluorooctane Surfactants in Lake Ontario B R Y A N B O U L A N G E R , † A A R O N M . P E C K , †,‡ JERALD L. SCHNOOR,† AND K E R I C . H O R N B U C K L E * ,† Department of Civil and Environmental Engineering, and IIHR-Hydroscience and Engineering, University of Iowa, Iowa City, Iowa 52240, and National Institute of Standards and Technology, Hollings Marine Laboratory, 331 Fort Johnson Road, Charleston, South Carolina 29412
Perfluorooctane surfactants have been reported in biota, water, and air samples worldwide. Despite these reports, the main environmental sources of these compounds remain undefined. To address this gap in knowledge, an annual lakewide mass budget of eight perfluorooctane surfactants was developed for Lake Ontario. To determine the atmospheric contribution to the mass budget, over-the-lake gasphase air concentrations for N-EtFOSE and N-EtFOSA and particulate-phase air concentrations for PFOS in any air sample are reported for the first time, with mean concentrations when present of 0.5 ( 0.32 (N-EtFOSE gasphase), 1.1 ( 0.9 (N-EtFOSA gas-phase), and 6.4 ( 3.3 (PFOS particulate-phase) pg/m3. The mass budget finds inflow from Lake Erie (14 361 ( 4489 kg Σperfluorooctane surfactants) and wastewater discharge (1762 ( 2697 kg Σperfluorooctane surfactants) to be the major sources, while outflow through the St. Lawrence River is the dominant loss mechanism (22 727 ( 7060 kg/year Σperfluorooctane surfactants). Using the mass budget data, the steady state and measured mean concentrations in the lake water are the same at the 95% confidence level.
Introduction Perfluorooctane surfactants are the building blocks of the commercial and industrial product lines used as surface protectors. This compound class is characterized by having a hydrophobic tail consisting of an eight-carbon aliphatic chain fully saturated with one of the strongest bonds in nature (C-F bond strength ) ∼110 kcal/mol) and a hydrophilic head (1). Because of the saturated carbon-fluorine bonds and the stability of the hydrophilic head, members of this compound class are valuable industrial and commercial surfactants, but are also environmentally stable. In 2000, 3M, a major manufacturer of perfluorooctane surfactants globally, produced an estimated three million kilograms of these chemicals for global industrial and commercial applications (2). In the United States approximately 41% of the sold product was coated onto paper and packaging products, 37% was impregnated into fabric, leather, and carpet goods, 10% was used as ingredients in industrial surfactants, additives, and coatings, and 3% was used in the production of aqueous fire-fighting foams (2). The widespread commercial and * Corresponding author phone: 319-384-0789; fax: 319-335-5660; e-mail:
[email protected]. † University of Iowa. ‡ National Institute of Standards and Technology. 74
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 1, 2005
industrial usage of perfluorooctane surfactant surface protectors has led to the accumulation of perfluorooctane sulfonate (PFOS) and perfluorooctanoic acid (PFOA), the end metabolic degradation products of the two major perfluorooctane surfactant classes, to measurable levels in marine biota and water throughout the globe and in the Great Lakes ecosystem (3-13). However, the cause of this accumulation is not well understood, because there is very little information pertaining to sources of these compounds to the environment. One possible source of these compounds to the Great Lakes is wastewater treatment plant discharge. The cleaning and care of surface-treated products (from clothing to carpets) by consumers and use in industrial processes are believed to release these compounds to municipal wastewater treatment systems. Additionally, treatment of landfill leachate by municipal water treatment works may also introduce significant amounts of these compounds to the environment from discarded treated products. In 1999, 3M studied the finished effluents of six wastewater treatment plants (WWTPs) for three perfluorinated compounds: PFOS, perfluorooctanoic acid (PFOA), and perfluorooctane sulfonamide (FOSA) (1). Four of these WWTPs were located in cities having a known source of perfluorinated compound manufacture or industrial use (test sites) and two were in cities that did not have known production or usage (controls). All six WWTP effluents had detectable concentrations of PFOS and PFOA ranging from 47.7 to 5290 ng/L and 41.2 to 2420 ng/L, respectively. FOSA was detected in the effluent of two WWTPs in test sites at concentrations of 56.1 to 85.3 ng/L, but was not found in the control cities (1). Identification of these compounds in finished effluent supports wastewater treatment plant discharge as a significant source to the Great Lakes. Another hypothesized source is gaseous and particulate atmospheric deposition. Volatilization of these compounds from production centers, treatment facilities, or from already distributed treated products may release these compounds into the atmosphere where they can be transported or deposited. To support this hypothesis, Martin et al. first reported gas-phase concentrations in air when they identified pg/m3 concentrations of N-methylperfluorooctane sulfonamidoethanol (N-MeFOSE), N-ethylperfluorooctane sulfonamidoethanol (N-EtFOSE), and N-ethylperfluorooctane sulfonamide (N-EtFOSA) in air samples from Toronto, Canada (14). Stock et al. recently reported pg/m3 concentrations of N-MeFOSE, N-EtFOSE, and N-EtFOSA in combined gas- and particulate-phase air samples from another six sites in North America, indicating the widespread nature of perfluorooctane sulfonamides in the atmosphere (15). However, both of these studies relied on quantification using gas chromatography mass spectrometry (GC/MS), which excludes analysis of a majority of perfluorooctane surfactants due to their low volatility. Additionally, these previous reports for air were all taken over land. For this paper, gas- and particulate-phase air samples taken over open water were analyzed using liquid chromatography mass spectrometry (LC/MS), allowing for the analysis of PFOS in air samples for the first time and a better approximation of atmospheric sources to the Great Lakes system. The primary objective of this work was to create a mass budget for perfluorooctane surfactants in Lake Ontario. Gaseous- and particulate-phase air sample concentrations reported in this paper were coupled with previously reported water and wastewater treatment plant effluent concentrations to construct the mass budget. Perfluorooctane surfactant 10.1021/es049044o CCC: $30.25
2005 American Chemical Society Published on Web 12/03/2004
FIGURE 1. Map of air transects A-H sampled in this study. Solid circles represent approximate water sampling locations collected at the same time as the air transects (16). mass inputs and outflows were estimated by assessing their main source loadings to the lake (gaseous and particulate atmospheric deposition, upstream inflow from Lake Erie, and discharge from wastewater treatment plants) and the known loss mechanisms affecting these compounds (outflow to St. Lawrence River and volatilization). Using these source inputs and loss mechanisms, the steady state concentrations of eight perfluorooctane surfactants in Lake Ontario are estimated. Steady state estimates are then compared to our previously reported Lake Ontario water concentrations (16). To our knowledge this work represents the first attempt to examine mass flows and identify sources of these compounds in any environmental system.
Methods Standards and Reagents. Standards of 2(N-ethylperfluorooctanesulfonamido) ethyl alcohol (N-EtFOSE, chemical purity 97.7%), 2-(N-ethylperfluorooctanesulfonamido) acetic acid (N-EtFOSAA, 98.6%), perfluorooctanesulfonylethylamide (N-EtFOSA, 99.3%), 2-(perfluorooctanesulfonamido) acetic acid (PFOSAA, 99.6%), perfluorooctane sulfonylamide (FOSA, 98.94%), perfluorooctane sulfonate potassium salt (PFOS, 86.9%), perfluorooctane sulfinate potassium salt (PFOSulfinate, 97.2%), and perfluorooctanoate postassium salt (PFOA, 95%) were provided by the 3M Company (St. Paul, MN). Compound structures are provided in the Supporting Information. HPLC grade methanol, dichloromethane, water, hexane, acetone, and ethyl acetate were purchased through Fisher Scientific. Air Sample Collection, Extraction, and Cleanup. Air sample transects were collected while onboard the R/V Lake Guardian from August 7-12, 2003 (Figure 1). Transects were designed to sample air over the lake influenced from urban areas (Detroit, MI and Toronto, ON), as well as non-urbaninfluenced air. The transects also provide a west to east distribution. Details of the air sample collection and extraction have been described elsewhere (17). Briefly, high-volume air samplers were used to collect eight air transects with replicate samples taken during the first and third transect. Over-water air sample volumes ranged from 95 to 378 m3 and were collected for 8-15 h. Air samples consisted of an operationally defined airborne particulate-phase, collected on a glass-fiber filter, and gas-phase, collected on XAD-2 resin in series. All XAD-2 resin was sequentially extracted for 24 h with methanol, acetone, dichloromethane, and hexane prior to sampling. Glass-fiber filters were combusted at 450
°C for at least 4 h and then desiccated and weighed prior to sampling. After sampling, the XAD-2 resin and glass-fiber filters were extracted separately with acetone/hexane for 24 h. The samples were then cleaned up using florisil columns and eluted with ethyl acetate. The extracts were reduced to ∼100 µL using N2 and then analyzed using LC/MS methods. Analytical Methods. Analytes were separated and quantified according to the method of Boulanger et al. (16). Briefly, analytes were chromatographically separated using an Agilent 1100 series HPLC/MSD system modified with low-deadvolume tubing. A 10-µL sample of each extract was injected onto a 150 × 2.1 mm (5 µm) Zorbax Extend C18 column (Agilent) with a 12.5 × 2.1 mm Narrow Bore C18 Guard (Agilent). The mobile phase consisted of (A) 2 mM ammonium acetate and (B) methanol. The flow rate was set at 0.3 mL/ min. The separation proceeded from initial conditions of 3% B held until 0.50 min, ramped to 95% B until 6.00 min, held until 8.50 min, and dropped to 3% B at 8.51 min. The total run time was 14 min with 1 min of postrun time between samples. The column temperature was held constant at 40 °C. Analytes were identified and quantified using selective ion monitoring (SIM) operating in negative electrospray ionization mode. Ions selected for monitoring in this study included the following: PFOA (m/z ) 413), PFOSulfinate (m/z ) 483), FOSA (m/z ) 498), PFOS (m/z ) 499), N-EtFOSA (m/z ) 526), PFOSAA (m/z ) 556), N-EtFOSAA (m/z ) 584), and N-EtFOSE alcohol-acetate adduct (m/z ) 630). Retention times for all analytes were between 9.5 and 12.5 min. The desolvation gas temperature was 300 °C at a flow rate of 8.0 L/min. The collision cell was maintained at 30 psig. Capillary voltage was set at 3500 V with a fragmentor voltage of 70 V. A dwell time of 0.8 s was used to monitor the ions. External calibration was used to quantify analyte values for LC/MS techniques using multicomponent standards. Each multicomponent standard contained every analyte at the same concentration. Five calibration standards (1 ng/mL, 5 ng/mL, 10 ng/mL, 50 ng/mL, and 100 ng/mL) were used to form the calibration curve for each compound. Quadratic curve fits were used for all analytes. QA/QC. Field blanks were defined as XAD-2 resin cartridges taken to the field, but not processed in the highvolume air samplers. These blanks serve as a blank for the resin itself, and did not have air pulled through them. Three blanks were used during the expedition. Samples were VOL. 39, NO. 1, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
75
TABLE 1. Recoveries for Florisil Cleanup and N2 Blowdown compound
% recovery from cleanup
% recovery from blowdown
PFOA PFOS PFOSulfinate PFOSAA N-EtFOSAA FOSA N-EtFOSE N-EtFOSA
6 79 63 2 0 99 106 102
93 116 54 13 61 84 91 88
TABLE 3. Comparison of Literature Values for N-EtFOSA and N-EtFOSE Reported in Gas-Phase Air Samplesa site
sample concentration size (pg/m3) year
group
ref.
2001 2001 2001 2002 2003 2003
Stock et al. Stock et al. Stock et al. Martin et al. this study this study
(15) (15) (15) (14)
2001 2001 2001 2002 2002 2003 2003
Stock et al. Stock et al. Stock et al. Martin et al. Martin et al. this study this study
(15) (15) (15) (14) (14)
N-EtFOSA Long Point, Canada Reno, Nevada Cleaves, Ohio Toronto, Canada Lake Erie Lake Ontario
3 3 3 4 5 3
Reno, Nevada Toronto, Canada Long Point, Canada Toronto, Canada Long Point, Canada Lake Erie Lake Ontario
3 3 3 4 2 5 3
∼10 ∼70 ∼45 14 (50) nd - 2.2 (40) nd - 1.5 (33)
N-EtFOSE
TABLE 2. Gas- and Particulate-Phase Perfluorooctane Surfactants in Air Sample Transectsa gaseous phase
particulate phase
transect ID
N-EtFOSE
N-EtFOSA
PFOS
A B C D E F G H
0.3 nd nd nd 1.0 0.4 0.6 nd
0.4 nd nd nd 2.2 1.5 nd nd
5.4 8.1 8.0 nd nd 2.5 nd nd
a
All concentrations are pg/m3; nd ) not detected in sample.
considered to contain the analytes if the mass was four times the mass found in the field blanks. To evaluate recoveries from the cleanup step, five 75-mm florisil columns were prepared and spiked with 400 ng of perfluorooctane surfactant standard containing each compound prepared in hexane. The columns were then eluted with 4 mL of ethyl acetate. Five additional florisil blank columns did not receive a spike, but were eluted with 4 mL of ethyl acetate. Aliquots of each column elution were analyzed. To evaluate recoveries from the blowdown step, five 4-mL volumes of ethyl acetate were spiked to 20 ng/mL with perfluorooctane surfactant standard containing each compound. These spiked samples were reduced to ∼100 µL using N2 and analyzed. Duplicate analysis of individual extracts as well as calibration check standards were run every fifteenth and twentieth sample, respectively. Instrument reproducibility was always within 5% for every duplicate and calibration check run.
Results QA/QC. Quality assurance was performed by spiked recovery studies of extraction methods and comparison of the sample masses to those in field blanks. The sample to blank mass ratio for all quantified samples in our study ranged from 4.1 to 25 for both gaseous-phase and particulate-phase samples. Analysis of the two replicate transects for gas-phase and particulate-phase yielded concentrations within 25%. The average recoveries from the cleanup and blowdown steps are summarized in Table 1. PFOS, PFOSulfinate, FOSA, N-EtFOSE, and N-EtFOSA had acceptable recoveries for both steps allowing for quantitative analysis of the air samples using LC/MS. PFOA, PFOSAA, and N-EtFOSAA had low recoveries. Therefore, retention on the florisil column prevented monitoring of these three analytes in our air samples. Air sample concentrations reported in this study were not corrected for loss during cleanup or blowdown. Air Concentrations. Only N-EtFOSE and N-EtFOSA were detected in over-the-lake gas-phase air samples (Table 2). Mean ( standard deviation N-EtFOSE and N-EtFOSA gas76
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 1, 2005
199 ∼80 ∼12 51-393 (100) 68, 85 (100) nd - 1.0 (40) nd - 0.6 (66)
a ( ) represents the percentage of positive detects within the sample size; ∼ represents values approximated from graphs within Stock et al. paper; nd ) not detected in sample.
phase concentrations when present for both lakes were 0.5 ( 0.32 and 1.1 ( 0.9 pg/m3, respectively. PFOS, PFOSulfinate, and FOSA were not detected in any gaseous-phase air samples. Only PFOS was detected in the particulate-phase air samples (Table 2). PFOS was detected in four out of the eight air samples with the mean concentration for both lakes when present of 6.4 ( 3.3 pg/m3. For both gas- and particulate-phase samples, analytes were detected only in samples taken close to Detroit and Toronto, indicating the importance of urban areas as a source for atmospheric perfluorooctane surfactants. Concentrations of N-EtFOSE and N-EtFOSA reported here are lower than gas-phase concentrations reported by other researchers (Table 3). Because past reports were sampled in urban environments, over-the-lake concentrations are expected to be lower. This trend has been observed with other persistent organic pollutants as well (18).
Mass Budget The overall contribution of perfluorooctane surfactants to the Great Lakes was evaluated through a mass budget prepared for Lake Ontario. Whenever possible, the mean ( standard deviation of input values were used in creating the budget. Input parameters where the standard deviation was unknown or not available were assumed constant. Error was propagated throughout the budget (19, 20). The daily fluxes into Lake Ontario for each compound are summarized in Figures 2 (mass inputs) and 3 (mass outflows).
Mass Fluxes Into Lake Ontario Inflow From Lake Erie. Flux into Lake Ontario due to Lake Erie’s inflow was calculated according to eq 1, where Finflow is determined using the 10-year mean total inflow into Lake Ontario from Lake Erie (Qinflow ) 4.94 × 108 m3/d) from the Lake Ontario Lakewide Management Plan (LaMP) (21) and the aqueous dissolved phase concentrations of these compounds in Lake Erie (Cinflow) for samples collected during the same sampling expedition (16). The water concentrations had mean ( standard deviation concentrations of PFOA, PFOS, PFOSulfinate, N-EtFOSAA, and FOSA equal to 36 ( 7.2, 31 ( 6.9, 5.0 ( 8.1, 7.5 ( 2.6, and 0.8 ( 0.3 ng/L, respectively.
Finflow ) Qinflow Cinflow
(1)
Gaseous Atmospheric Deposition. Gross atmospheric deposition influx (Fd) to Lake Ontario was estimated using
FIGURE 3. Estimated daily mass fluxes of perfluorooctane surfactants (kg/d) out of Lake Ontario: error bars represent the estimated daily flux for the mean concentration ( one standard deviation; nd ) compounds not detected in samples.
FIGURE 2. Estimated daily mass fluxes of perfluorooctane surfactants (kg/d) into Lake Ontario: error bars represent the estimated daily flux for the mean concentration ( one standard deviation; nr ) compounds that may be present, but were not recovered during analysis; nd ) compounds that are recovered during analysis, but not detected in samples; na ) compounds not available from literature sources. eq 2, where kol is the overall mass transfer coefficient, Ca is the gas-phase concentration over the lake (this study), H is the Henry’s Law constant (unitless) (1), and A is the surface area of Lake Ontario (1.9 × 1010 m2) (21). The overall mass transfer coefficient was determined as described elsewhere (22, 23) assuming 20 °C and a 5 m/s wind speed. The value of kol ranged from 5.15 × 10-5 m/d for PFOS to 0.4 m/d for N-EtFOSE.
Fd )
()
Ca k A H ol
(2)
Particle Atmospheric Deposition. Gross atmospheric deposition influx (Fpd) to Lake Ontario was estimated using eq 3, where υs is the settling velocity for particles over the lake, Cp is the measured particulate concentration, and A is the surface area of the lake. The settling velocity is approximated at 0.15 m/s as described by others (24).
Fpd ) υsCpA
(3)
Wastewater Treatment Discharge. The input to the lake from wastewater treatment plants (FWWTP) was determined from the total volumetric flow rate of wastewater into Lake Ontario (QWWTP) and perfluorooctane surfactant concentrations in wastewater treatment plant effluent (CWWTP) as described in eq 4.
FWWTP ) QWWTPCWWTP
(4)
Wastewater treatment plant effluent is discharged into Lake Ontario from both the United States and Canada. The average flow of wastewater treatment plant effluent into Lake Ontario over a 10 year period, as reported in the Lake Ontario LaMP, including both direct discharge into the lake and discharge into tributaries that feed into the lake, was 2.74 × 106 m3/day (21). Concentrations and standard deviations used in the model for wastewater effluents were reported by 3M for FOSA, PFOS, and PFOA and were 37 ( 26, 1140 ( 1823, and 549 ( 840 ng/L, respectively (1). Total Annual Estimated Mass Influx. The mass inflow from Lake Erie is the largest source to Lake Ontario and contributes an estimated 6438 ( 1287 kg of PFOA, 5544 ( 1234 kg of PFOS, 894 ( 1448 kg of PFOSulfinate, 1341 ( 465 kg of N-EtFOSAA, and 143 ( 53 kg of FOSA to Lake Ontario per year due to inflow alone. While the hydrologic inflow may vary between 5 and 20% on an annual basis, influx from the Niagara River will govern the concentration of perfluorooctane surfactants in Lake Ontario even if the hydrologic variability is included because the volumetric flow from this input is much larger than all other flows to the lake. Assuming the reported perfluorooctane surfactant concentrations in WWTP effluent presented by 3M are representative of actual inputs to the lake, 548 ( 841 kg of PFOA, 37 ( 26 kg of FOSA, and 1141 ( 1825 kg of PFOS are discharged into the lake each year from wastewater treatment plants. The high degree of uncertainty associated with these estimations occurs due to the limited number of 3M samples studied. Additionally, the 3M study also did not examine WWTP effluents discharging to the Great Lakes system. Therefore, additional research is needed to accurately determine the contributions of WWTP effluents in the Great Lakes region. On an annual basis gas- and particulate-phase atmospheric transfer into the lake is negligible (
