Environ. Sci. Technol. 2007, 41, 8248–8255
Semivolatile Fluorinated Organic Compounds in Asian and Western U.S. Air Masses ARKADIUSZ M. PIEKARZ,† TOBY PRIMBS,† JENNIFER A. FIELD,‡ D O U G L A S F . B A R O F S K Y , †,‡ A N D S T A C I S I M O N I C H * ,†,‡ Department of Chemistry, Oregon State University and Department of Environmental and Molecular Toxicology, Corvallis, Oregon
Received June 7, 2007. Revised manuscript received September 26, 2007. Accepted October 9, 2007.
Semivolatile fluorinated organic compounds (FOCs) were measured in archived air sample extracts collected from Hedo Station Observatory (HSO) on Okinawa, Japan and Mount Bachelor Observatory (MBO), Oregon U.S. during the springs of 2004 (MBO and HSO) and 2006 (MBO). Fluorotelomer alcohols (FTOHs) were measured in both Asian and western U.S. air masses, though western U.S. air masses had significantly higher concentrations. Concentrations of fluorotelomer olefins in Asian air masses and 8:2 fluorotelomer acrylate in U.S. air masses were reported for the first time. N-ethyl perfluorooctane sulfonamide, N-methyl perfluorooctane sulfonamido ethanol, and N-ethyl perfluorooctane sulfonamido ethanol were also measured in Asian and western U.S. air masses but less frequently than FTOHs. The atmospheric sources and fate of FTOHs were investigated by estimating their atmospheric residence times, calculating FTOH concentration ratios, investigating FTOH correlations with nonfluorinated semivolatile organic compound concentrations, and determining air mass back trajectories. Estimated atmospheric residence times for 6:2 FTOH, 8:2 FTOH, and 10:2 FTOH were 50, 80, and 70 d, respectively, and the average concentration ratio of 6:2 FTOH to 8:2 FTOH to 10:2 FTOH at MBO in 2006 was 1.0 to 5.0 to 2.5. The relative order of these atmospheric residence times may explain the observed enhancement of 8:2 FTOH and 10:2 FTOH (relative to 6:2 FTOH) at MBO compared to North American indoor air (FTOH average ratio of 1.0 to 2.0 to 1.0). FTOH concentrations in western U.S. air masses were positively correlated (p < 0.05) with gas-phase polycyclic aromatic hydrocarbon and polychlorinated biphenyl concentrations and negatively correlated (p < 0.05) with agricultural pesticide concentrations. In comparison to western U.S. air masses, transPacific air masses did not contain elevated concentrations of these compounds, whereas lower boundary layer air masses that passed over urban areas of the western U.S. did. This suggests that semivolatile FOCs are emitted from urban areas in the western U.S.
* Corresponding author phone: (541) 737-9194; fax: (541)7370497; e-mail:
[email protected]. † Department of Chemistry. ‡ Department of Environmental and Molecular Toxicology. 8248
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 24, 2007
Introduction Perfluorooctane sulfonate (PFOS) and other long-chain perfluorinated carboxylic acids (PFCAs) have been detected in mammals across the North American Arctic (1–4), dolphins in the Atlantic Ocean (5), seawater samples from the Pacific and Atlantic Oceans (6), and in North American precipitation (7, 8). Due to their ionic nature, PFOS and PFCAs are relatively nonvolatile and the mechanism by which they undergo longrange atmospheric and oceanic transport to remote locations is not well understood (9–11). Semivolatile fluorotelomer alcohols (FTOHs), fluorotelomer olefins (Ftenes), N-methyl perfluorooctane sulfonamidoethanol (N-MeFOSE), N-ethyl perfluorooctane sulfonamidoethanol (N-EtFOSE), and N-ethyl perfluorooctane sulfonamide (N-EtFOSA) have been reported in the atmosphere near urban areas in the United States, Germany, and Canada, as well as in remote areas of the North Atlantic Ocean, the Canadian Archipelago, and the western ocean coast of Africa (12–18). FTOHs, N-MeFOSE, N-EtFOSE, and N-EtFOSA have also been reported in indoor air (18–20), and gas-particle partitioning of FTOHs, N-MeFOSE, and N-EtFOSE has been observed (12-16), (18-20). Smog chamber studies indicate that regional and longrange atmospheric transport of FTOHs may be due to their relatively slow rate of reaction with hydroxyl radical in the atmosphere (21). The atmospheric lifetimes of FTOHs and N-methylperfluorobutanesulfonamide (N-MeFBSA), estimated from these smog chamber studies, were at least 20 d, suggesting that, regardless of chain length, FTOHs and perfluorosulfonamides degrade into a series of homologue perfluorosulfonates and PFCAs (21–25)]. Snow samples from the high Arctic also indicate that atmospheric transport of these FOCs may be the source of PFOS and PFCAs found in remote locations (26). In the present study, semivolatile FOCs were quantified in Asian and western U.S. using archived high-volume air samples collected from two remote locations: Hedo Station Observatory on Okinawa, Japan (HSO) and Mount Bachelor Observatory (MBO), Oregon, during the springs of 2004 (HSO and MBO) and 2006 (MBO). The types and concentrations of FOCs in Asian and western U.S. air masses are reported. In the comprehensive data set from MBO in 2006, the origin of the fluorinated organic chemicals were traced to urban source regions using FTOH concentration ratios, FTOH correlations with nonfluorinated semivolatile organic compound (SOC) concentrations, and air mass back trajectories (27). Finally, the atmospheric residence times of individual FTOHs were estimated using a trace gas variability method developed by Junge (28).
Experimental Details Chemicals, Sampling Media, and Materials. Table SI.1 in the Supporting Information shows the chemical names, acronyms, structures, vendors, and purities of each target analyte investigated in this research. The FOCs detected in this study included perfluorohexane ethanol (6:2 FTOH), perfluoroctane ethanol (8:2 FTOH), perfluorodecane ethanol (10:2 FTOH), perfluorodecane ethylene (10:2 Ftene), perfluurododecane ethylene (12:2 Ftene), perfluoroctane ethyl acrylate (8:2 FtAc), N-methylperfluoroocatne sulfonamido ethanol (N-MeFOSE), N-ethylperfluorooctane sulfonamido ethanol (N-EtFOSE), and N-ethylperfluorooctane sulfonamide (N-EtFOSA). Perfluoroheptane methanol (7:1 fluorinated alcohol or PDFO was used as the internal standard (MBO and HSO 2004 samples) or surrogate (MBO 2006 10.1021/es0713678 CCC: $37.00
2007 American Chemical Society
Published on Web 11/15/2007
samples). Further details on the FOCs measured in this study are available in the Supporting Information. Sampling Sites and Sample Collection. Eighteen, 24 h air samples (600–800 m3) were collected from March 22 through May 2, 2004 at the Hedo Station Observatory (HSO), located 60 m above sea level (masl) on the northwest coast of Okinawa, Japan (26.8°N, 128.2°E) (29). Eight 24 h air samples (600–800 m3) were collected from April 22 through July 21, 2004 and thirty-four 24 h air samples (600–800 m3) were collected from April 3 through May 13, 2006 at Mount Bachelor Observatory (MBO). MBO is located on a mountaintop, 180 km east of the Pacific Ocean, in the Cascade region of central Oregon (43.98°N 121.69°W 2700 masl) and is a relatively new site for studying trans-Pacific atmospheric transport [ (30, 31)]. Sampling start dates, air sample volumes, and other meteorological data for HSO 2004 are given in Primbs et al. (29), and the MBO 2004 and 2006 sampling data are shown in Tables SI.2A and SI.2B. To the best of our knowledge, there are no known fluorochemical manufacturing sites near HSO or MBO. Details of the modified high volume air samplers are described in Primbs et al. (29). Briefly, the high volume air sampler comprised two quartz-fiber filters (QFFs), in series, to collect particulate matter (first QFF) and estimate adsorption of gas-phase FOCs (second QFF), followed by a polyurethane foam-styrene-divinylbenzene-resin-polyurethanefoam (PUF-XAD-PUF) sandwich to collect gas phase compounds. Further detail about the preparation of QFFs, PUFs, and XAD-2 resin, as well as the storage and transport of samples, can be found in Primbs et al. (29) . Previous research groups have used similar high volume air sampling approaches to measure FOCs in outdoor air (12-19, 32). Sample Extraction and Concentration. Pressurized liquid extraction (PLE) and extract concentration were performed using Accelerated Solvent Extraction (ASE-300) (Dionex, California) and a Turbovap II (Caliper Life Sciences, MA) units, respectively. The QFFs (representing the particulatephase fraction) and the PUF and XAD-2 (representing the gas-phase fraction) were extracted and analyzed separately. The XAD-2 resin and QFFs were extracted with ethyl acetate, followed by dichloromethane, while the polyurethane foam (PUF) was extracted once using a 75:25 hexane:acetone mixture (29). The extracts were concentrated to 600 µL using the Turbovap II, and the PUF extract was solvent exchanged to ethyl acetate, followed by gentle blow down to 300 µL using ultrapure nitrogen (29). PDFO served as the internal standard for the HSO and MBO 2004 samples and was spiked into each extract vial just prior to GC injection. For the MBO 2006 samples, the PDFO served as a surrogate and was spiked prior to extraction, onto the sampling media. This variation in procedure was dictated by the fact that the HSO and MBO 2004 samples were extracted before the present study was conceived, whereas the MBO 2006 samples were extracted afterward. Thus, the concentrations at HSO and MBO in 2004 were not recovery-corrected and should be compared qualitatively, whereas concentrations at MBO in 2006 were extraction recovery-corrected based on the recovery of PDFO. GC/MS Analysis and Limit of Detection. The fluorochemicals were analyzed via gas chromatography–mass spectrometry (GC/MS). A JEOL GC-Mate II magnetic sector mass spectrometer, operating in a positive chemical ionization (PCI) mode, was used for quantitative analyses, whereas an Agilent 6890 GC/Agilent 5973N MS detector, operating in a negative chemical ionization (NCI) mode, was used to confirm the identities of the target analytes. Details of the instrumentation and analysis are available in the Supporting Information. Chromatographic separation of nonolefinic compounds was performed on an EC-Wax column (Altech, 30 m × 0.25 mm × 0.25µm); fluorinated olefins were separated
on a less polar DB-5MS Column (J&W Scientific, 30 m × 0.25 mm × 0.25 µm). Table SI.3 lists the retention times and m/z ions monitored for each analyte in PCI/NCI mode. A National Institute for Occupational Safety and Health (NIOSH) procedure (33) was used to determine the instrumental limit of detection (LOD) and instrumental limit of quantitation (LOQ) for each target analyte. Further detail about the NIOSH LOD procedure is provided in the Supporting Information. Matrix-specific estimated detection limits (EDLs), using EPA method 8280A, were calculated for the analytes in the air samples and ranged from 0.3 pg/m3 (12:2 Ftene and 6:2 FTOH) to 1 pg/m3 (N-MeFOSE and N-EtFOSE) (34)]. Table SI.4 shows instrumental LODs for each combination of GC column, mass spectrometer, and ionization mode used during GC/MS analysis and method EDLs. All air concentrations assigned as less than LOQ or LOD in this work are based on EDLs determined by EPA method 8280A and indicate the matrix-specific (air) limits of quantitation and detection respectively. Method Recovery Experiments. The analytical method was validated for the fluorochemicals listed in Table SI.1 by performing spike and recovery experiments on the PUF, XAD-2 resin, and QFF, respectively. Absolute recoveries were determined in separate experiments by spiking 30 µL of a 10 ng/µL solution of each target analyte before the extraction step (to assess losses during extraction and extract concentration, n ) 3) and before the concentration step (to assess loss during the use of the TurboVap II, n ) 2). This approach made it possible to determine the extents to which extraction and extract concentration contributed to analyte loss. In both sets of experiments, the internal standard, PDFO, was spiked into the GC vial prior to injection. Air Mass Back Trajectories. The source regions of the air masses sampled were determined using the hybrid singleparticle lagrangian integrated trajectory (HYSPLIT) 4.0 model, FNL (191 km resolution) (27). For HSO, 4-d back trajectories were calculated, and source region impact factors (SRIFs) were determined for each 24 h sample. A SRIF describes the percentage of time that an air parcel spends in a particular source region before arriving at the sampling location. Source regions for the HSO 2004 samples included Japan, North and South Korea, and the Pacific Ocean (which includes the potential influence of the island of Okinawa). Details of the calculation of SRIFs and the results for the HSO Spring 2004 sampling campaign for nonfluorinated SOCs are summarized by Primbs et al. (29). For MBO 2006, three-day back trajectories were calculated using the HYSPLIT 4.0 model, EDAS (40 km resolution). Trajectories over a 0.2° × 0.2° box centered on MBO were calculated every 3 h, over a 24 h period, at 0.04° latitude and longitude increments, and at three elevations (1300, 1500, and 1700 m above model ground level); 675 trajectories were calculated for each 24 h sample. Atmospheric Residence Time. From measured concentrations of FTOHs at MBO in 2006, the Junge method was used to calculate the atmospheric residence times (28). This method, which uses the variability in the concentration of tropospheric trace gases to estimate the atmospheric residence time, has been used previously for nonfluorinated SOCs like polychlorinated biphenyls (PCBs) (35–38). According to Junge, a trace gas, whose measured concentration has a relative standard deviation σ, has a residence time given by the equation Tσ ) 0.14, where T is the atmospheric residence time in years.
Results and Discussion Data Quality Assurance. The sampling campaigns at HSO and MBO included laboratory blanks, travel blanks, and field blanks. Each type of blank included a QFF, PUF, and XAD-2 resin. Field blanks were passively collected by placing the VOL. 41, NO. 24, 2007 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8249
sampling media in the air sampler for 24 h on days when high volume sampling was not conducted. No analytes were detected in the laboratory, and travel blanks from either the HSO or MBO sites, with the exception of one of the MBO 2004 travel blanks, for which 8:2 FTOH was detected at less than the LOQ in the XAD-2 resin and QFF fractions. The HSO and MBO 2004 sampling campaigns included three and two field blanks, respectively. One of the field blanks from HSO contained 6:2 FTOH and 8:2 FTOH at less than the LOQ in the XAD-2 fraction. In one of the MBO 2004 field blanks, the XAD-2 fraction had measurable concentrations of 6:2 FTOH, 8:2 FTOH, and 10:2 FTOH (less than 2 pg/m3) and measurable concentrations of 8:2 FTOH, 10:2 FTOH, MeFOSE, and EtFOSE in the QFF fraction (less than 3 pg/ m3). This anomalous occurrence was most likely the result of the blank media having been passively exposed to air for a week, instead of a day, before being removed from the sampler due to bad weather. In contrast, all field blanks and samples from the HSO site were removed from the sampler after 24 h. Four, 24 h field blanks were collected during the course of the sampling at MBO in 2006. For these four field blanks, 8:2 FTOH was detected from below the LOQ to ∼3 pg/m3; 10:2 FTOH was detected from below the LOQ to ∼2 pg/m3; N-MeFOSE was detected at the LOQ in one field blank; and N-EtFOSA was detected at ∼1 pg/m3 in one field blank. No other analytes listed in Table SI.1 were detected in any of the field blanks collected at the HSO and MBO sites. Field blank concentrations were not subtracted from the measured sample concentrations. Approximately 30% of the samples collected at the HSO and MBO sites were randomly selected, and the back PUFs from the samples were analyzed to determine analyte breakthrough and back QFFs were analyzed to determine the sorption of gas-phase analytes on the filter. No analytes were detected in these fractions. Thus, the concentrations of analytes in the gas-phase determined from samples collected at the HSO and MBO sites are not likely to be underestimated due to analyte breakthrough, nor are the concentrations of the analytes measured in the particulate phase likely to be overestimated as a result of sorption to the QFF. Fluorochemical Recoveries. Known amounts of the FOCs were spiked onto the sampling media just prior to extraction and PDFO was spiked just prior to analysis in order to determine FOC extraction and concentration recoveries. The average recoveries of FOCs ranged from 0% (8:2 Ftene) to 74% (MeFOSE in the XAD-2 fractions) (Table SI.5). The majority of FOC loss was due to volatilization during the extract concentration step. As a result of the analyte recoveries being less than 100% and the concentrations measured at HSO and MBO in 2004 not being corrected for PDFO recovery, the concentrations reported for HSO and MBO in 2004 likely underestimate the actual concentrations in the atmosphere by a factor of 1.4 to 16, depending on the analyte (Table SI.5). While the FOC concentrations measured at MBO in 2006 account for PDFO loss due to extraction and extract concentration, these recovery-corrected FOC concentration data may overestimate the concentrations of N-MeFOSE, N-EtFOSE, and N-EtFOSA in MBO 2006 data due to the volatility of PDFO. Higher recoveries of Ftenes, FTOHs, and fluorinated sulfonamides have been reported in the literature for analytical methods specifically designed to optimize the recovery of these compounds (12-15, 18, 19, 32). However, Jahnke et al. reported recoveries of the internal standards 6:2 FTOH [M+4] and 8:2 FTOH [M+4] to be 12-60% and 24-80%, respectively, and the recovery of NMeFOSA [M+3] and NEtFOSA [M+5] to range from 54 to 155% (15). Shoeib et al. reported extraction-corrected recoveries of 60 and 47% 8250
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 41, NO. 24, 2007
for MeFOSE and EtFOSE, respectively (19). Finally, the first method papers on FOC measurements in air did not include method recoveries in the validation description of the method (16, 17). These examples indicate that the recoveries reported in this paper are representative of other methods used to collect FOCs; moreover, there was no gross overestimation of environmental concentrations in this study. In this study, the archived extracts from previously collected samples were analyzed for FTOHs, fluorotelomer olefins, and fluorinated sulfonamides; consequently, no control could be exercised over the recoveries of the fluorochemicals from these samples. The methods used in the present study (high-volume air sampling with two front QFFs followed by PUF-XAD-PUF sandwich, solvent extraction and extract concentration by solvent evaporation) are, however, similar to those used in these earlier studies that had been designed to specifically sample, extract, and concentrate FTOHs and fluorinated sulfonamides (12-19, 32). Moreover, the MBO 2006 samples provided a unique opportunity to investigate correlations between FOCs and nonfluorinated SOC concentrations. Air Concentrations of FOCs. The total concentrations (Σgas and particle) of fluorochemical analytes measured at the HSO 2004, MBO 2004, and MBO 2006 sampling sites are given in Figure 1. Okinawa, Japan (HSO) 2004. Fluorotelomer olefins (Ftenes) were only detected in Asian air masses at HSO. The 10:2 and 12:2 Ftenes were measured only in the gas phase at concentrations as high as 1.6 and 2.2 pg/m3 and with frequencies of detection above the LOD of 61 and 22%, respectively. However, only two HSO air samples contained these two Ftenes at concentrations above the LOQ (Figure 1). A possible explanation for the presence of 10:2 Ftene and 12:2 Ftene, which have high vapor pressures, in Asian air, is the use of Ftenes at three sites in mainland Japan to synthesize fluorotelomer acrylate monomer or ammonium perfluorononanoate (APFN) (11). The primary olefin used in the synthesis of APFN is 8:2 Ftene; however, this compound, which almost coelutes with ethyl acetate on the 30 m DB5MS GC column, is too volatile to be recovered by the method used in this study (Table S.5) (11). Another possible explanation for the presence of Ftenes in the Asian air masses could be sources in China. Previously calculated SRIFs for the two sample dates that yielded quantifiable concentrations of Ftenes (April 1 and April 4, 2004) indicate large source region contributions (greater than 40%) from China, and to a lesser degree, from the Pacific Ocean (29). To further assess the magnitude and significance of Ftenes in the atmosphere, methods must be found for increasing their recoveries. Barber et al. described a slightly improved method for Ftenes using Carboxen 569 and subsequent thermal desorption (13). At HSO, the total concentrations of FTOHs (Σ(6:2 FTOH, 8:2 FTOH, and 10:2 FTOH)) in the gas phase ranged from
