Anal. Chem. 1998, 70, 4074-4080
Extraction and Analysis of Trifluoroacetic Acid in Environmental Waters Chad E. Wujcik,* Thomas M. Cahill, and James N. Seiber
Center for Environmental Sciences and Engineering and Department of Environmental and Resource Sciences, University of Nevada, Reno, Nevada 89557
Trifluoroacetic acid (TFA), a mildly phytotoxic compound, is a stable atmospheric breakdown product of HFC-134a, HCFC-123, and HCFC-124. An extraction and analytical method has been developed for the routine analysis of low ppt levels of TFA in aqueous samples. TFA can be quantitatively recovered from most environmental waters by an extraction procedure using a commercial anionexchange disk. In saline samples (conductivity >620 µS), where the presence of competing anions interfered with recovery, a liquid-liquid extraction cleanup was necessary. After extraction of TFA from water, the dried disk was placed in a headspace vial containing 10% sulfuric acid in methanol and the vial sealed and then vortexed for 30 s. The sulfuric acid-methanol solution extracts trifluoroacetate anion (TFA) from the anion-exchange matrix and, when heated, quantitatively converts it to the methyl ester, which is then analyzed by automated headspace gas chromatography using electron capture or mass spectrometry detection. Several environmental samples in addition to laboratory spike solutions were successfully extracted and analyzed with this technique. Recoveries averaged 108.2% for reagent water spiked at levels from 53 to 2110 ng/L with relative standard deviations ranging from 0.3 to 8.4%. The instrument’s limit of detection for TFA standard was 3.3 ng. The limit of quantitation for the extraction and analytical technique was 36 ng/L. Three water samples can be prepared for automated analysis in 20 min using this technique. The breakdown of three chlorofluorocarbon (CFC) replacement compounds, namely, HFC-134a, HCFC-123, and HCFC-124, results in the formation of trifluoroacetic acid (TFA).1-4 As a strong acid with a pKa of 0.47 and Henry’s law constant value of (8.9 ( 2) × 103 mol/kg‚atm,5 TFA will favor partitioning into aqueous environmental compartments and remain there due to (1) AFEAS (Alternative Fluorocarbons Environmental Acceptability Study), Research Summary, 1994. (2) McCulloch, A., AFEAS Workshop on the Environmental Fate of TFA, Miami Beach, Florida, March 3-4, 1994. (3) Wallington, T. J.; Schneider, W. F.; Worsnop, D. R.; Neilsen, O. J.; Sehested, J.; Debruyn, W. J.; Shorter, J. A. Environ. Sci. Technol. 1994, 28, 320A326A. (4) Wallington, T. J.; Hurley, M. D.; Fracheboud, J. M.; Orlando, J. J.; Tyndall, G. S.; Sehested, J.; Mogelberg, T. E.; Nielsen, O. J. J. Phys. Chem. 1996, 100, 18116-18122. (5) Bowden, D. J.; Clegg, S. L.; Brimblecombe, P. Chemosphere 1996, 32, 405420.
4074 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
the high solubility of the dissociated anion. Since most abiotic and biotic degradation pathways are relatively ineffective,6-8 TFA may accumulate in seasonal and terminal surface water systems.9 Although TFA is weakly phytotoxic,10 harmful levels could be reached over time in surface water systems that are characterized by little or no outflow. To conduct routine monitoring of environmental levels of TFA, an accurate, safe, and simple analytical method is needed. Current methods require extensive operator skill and involvement, the handling of hazardous derivatization reagents, or are lacking in sensitivity.11-13 For example, Frank et al.11 have developed a multistep extraction followed by an overnight derivatization technique using (pentafluorophenyl)diazoethane esterification. The reagents are carcinogenic and/or explosive. Ion chromatography,13 although quite useful for concentrations above high micrograms per liter, does not provide the sensitivity necessary to determine TFA at nanogram per liter environmental levels. The technique presented in this article was designed to reduce these problems by modifying and combining elements from existing methods. EXPERIMENTAL SECTION Materials. Extraction of aqueous samples used a 47-mm 3M anion-exchange SR Empore disk (3M, Minneapolis, MN). A Waters extraction assembly for a 47-mm filter, consisting of a 250300-mL graduated reservoir, ground glass fritted funnel, and clamp (Waters Corp., Milford, MA) was used to hold the Empore disk. A large rubber stopper, 4-L vacuum flask, and vacuum adjusting aperture valves completed the extraction assembly. Environmental samples were filtered with Whatman GF/F-type filters (Maidstone, England). Two variable-volume pipettors (Oxford Labware, St. Louis, MO) with ranges of 1-1000 and 1-5000 µL were used for the addition of standards, spikes, and other solutions. At least (6) Emptage, M. AFEAS Workshop on Decomposition of TFA in the Environment, Washington, DC, February 8-9, 1994. (7) Emptage, M.; Tabinowski, J.; Odom, J. M. Environ. Sci. Technol. 1997, 31, 732-734. (8) Visscher, P. T.; Culbertson, C. W.; Oremland, R. S.; Nature 1994, 369, 729-731. (9) Sefchick, J.; Skorupa, J.; Schwarzbach, S. AFEAS Workshop on the Environmental Fate of TFA, Miami Beach, FL, March 3-4, 1994. (10) Thompson, R. AFEAS Workshop on the Environmental Fate of TFA, Miami Beach, FL, March 3-4, 1994. (11) Frank, H.; Renschen, D.; Klein, A.; Scholl, H. J. High Resolut. Chromatogr. 1995, 18, 83-88. (12) Zehavi, D.; Seiber, J. N. Anal. Chem. 1995, 68, 3450-3459. (13) Hakins, D. C.; Kharasch, E. D. J. Chromatogr. B 1997, 692, 413-418. S0003-2700(98)00212-1 CCC: $15.00
© 1998 American Chemical Society Published on Web 08/27/1998
three 500-mL separatory funnels with Nalgene stopcocks and lids are necessary for each liquid-liquid extraction. Cleaning Procedure. All glassware was carefully cleaned to minimize residual TFA. The procedure involved cleaning with hot tap water several times with and then without soap, followed by five rinses in deionized water, and thorough rinsing with regent water and HPLC grade acetone. The glassware was either rapidly air-dried to remove acetone or, preferably, oven-baked at 150 °C for 3 h. Hot glassware was immediately covered with aluminum foil and sealed to prevent condensation and contamination. Blanks were continuously run with each batch of samples throughout this study to determine trace levels of TFA that could not be removed.11,12 Reagents. Sodium TFA (99+%) used in standards and spiking solutions was purchased from Sigma Chemical Co. (St. Louis, MO). Methyl trifluoroacetate (MTFA, 99%), used for standards, was purchased from Aldrich (Milwaukee, WI). Acetone (pesticide grade), methanol (pesticide grade), sulfuric acid (95-98%), and sodium hydroxide (99%) were obtained from Fisher Scientific (Pittsburgh, PA). High-purity reagent water of 18.0 MΩ‚cm specific resistance was obtained from a Barnstead Nanopure system (Dubuque, IA). Both granular sodium sulfate (analytical grade) and anhydrous diethyl ether (GR, 99%, containing peroxide inhibitor) were from EM Science (Gibbstown, NJ). Standards and Solutions. Standards of 1-500 ng/mL of both TFA and MTFA were prepared in methanol. Spike solutions of TFA in water were made at 9.3 and 462.8 ng/mL. A derivatization solution of 10% sulfuric acid in methanol and a conditioning solution of 1.0 M sodium hydroxide in water were made in 1-L volumes and replaced every 8 days of use. Sodium chloride solutions of 200, 400, 1000, and 2000 mg/L in high-purity reagent water were used for disk recovery experiments and conductivity calibration. Solutions of sodium sulfate at 500, 1000, and 2000 mg/L were also used for the disk recovery trials. Environmental Samples. A variety of environmental surface water samples were obtained during summer 1997, in California and Nevada to test recovery efficiency. These samples included a pristine mountain lake (Lake Tahoe, CA), the lower end of the Truckee River, NV, downstream from the city of Reno and its wastewater treatment plant, and more saline systems such as Pyramid Lake, NV, Mono Lake, CA, and the Pacific Ocean at San Francisco, CA. Two identical samples from Woodward Lake in Fresno, CA obtained on 1/4/96, one of which was sealed and not previously analyzed, were used for comparing this method with the evaporative concentration technique developed by Zehavi and Seiber12 and samples analyzed by Wujcik et al.14 The rainwater sample was a combination of two rainwater collections in the Central Valley at Selma, CA, during January and February 1996, to make a single consistent sample of 4 L in volume. A fogwater sample taken on 2/4/95 at 12:00-9:00 a.m. had been analyzed by Wujcik et. al.,14 but had been opened to the air on several occasions since analysis. These precipitation samples represent highly polluted samples with elevated particulate levels and concentrations of organics that could interfere in the analysis. Chlorinated tap water obtained from the Bureau of Mines Building on the University of Nevada campus on 8/27/97 was also used to (14) Wujcik, C. E.; Zehavi, D.; Seiber, J. N. Chemosphere 1988, 36, 1233-1245.
test the method. Samples predating 1997 were stored in amber glass bottles with PTFE-lined lids. More recent samples (August 1997 to present) were stored in amber bottles with aluminum foil lining the PTFE lid. All samples were stored at 5 °C. All samples were rapidly filtered at 5 °C with a GF/F-type filter to remove particulates. All efforts were made to reduce the exposure of aqueous samples to the ambient air. Conductivity Measurements. A set of three standards of NaCl in reagent water, in addition to a blank, was used to determine the conductivity value at which effective disk recovery diminished. These standards, consisting of 200, 400, and 1000 mg/L NaCl, were calibrated and measured by a H18033 Hanna Instruments conductivity meter (Woonsocket, RI). Environmental sample conductivity measurements were performed at 20 °C following calibration with the NaCl standards and a set of premade conductivity solutions from Orion (Beverly, MA). Derivatization Yield. The derivatization of TFA in 10% sulfuric acid in methanol was compared to the respective MTFA solution for determination of yield. Derivatization was performed in 22-mL headspace vials capped with PTFE-coated butyl rubber septa (Perkin-Elmer, Norwalk, CT) with 4 mL of 10% sulfuric acid in methanol. The 10% sulfuric acid in methanol solution was made by adding 400 µL of 18 M sulfuric acid to a cooled headspace vial, followed by 3600 µL of the appropriate methanol blank, MTFA standard in methanol, or TFA standard in methanol, and then capping immediately. Derivatization was carried out at 50 °C during the 1-h thermostating step on the automated headspace sampler. The derivatization of TFA extracted from 400 mL of spiked reagent water using a 3M anion-exchange Empore disk was also performed and compared to the above 4-mL standard solutions of 10% sulfuric acid in methanol. All blanks and standards for both TFA and MTFA were derivatized in triplicate. Anion-Exchange Procedure. Conditioning/Sample Extraction. The anion-exchange SR Empore disk was centered in a 300mL Waters filtration assembly and wet with ∼2 mL of acetone. The upper reservoir was then clamped in place and vacuum applied slowly through the disk until dry. An additional 15 mL of acetone was dispensed into the reservoir directly on the disk, followed by vacuum-drying. The vacuum was then closed and 15 mL of methanol added into the reservoir. The methanol was then slowly pulled through with vacuum until ∼5 mm above the surface of the disk. This was then followed by 15 mL of reagent water, 15 mL of 1 M sodium hydroxide, and another 15 mL of reagent water, with each being added in succession to the 5-mm level. The disk was not allowed to go dry during the conditioning or sample addition. Once the final reagent water reached 5 mm, the appropriate blank, spiked standard, or environmental sample was slowly added to the reservoir. For greatest reproducibility, samples were pulled through at a rate of ∼30-45 mL/min. Once the aqueous sample had reached the 5-mm level, the reservoir and disk were rinsed with three 15-mL aliquots of methanol and then allowed to dry. Disk Extraction/Derivatization. After the blank or sample was extracted the dried Empore disk was removed from the extraction assembly with clean, powderless latex gloves, cut into small pieces with a scissors, and placed in the 22-mL headspace vial. To this vial was added 4 mL of 10% sulfuric acid in methanol solution. The vial was capped immediately, vortexed for 30 s, and placed Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4075
on the headspace autosampler for analysis. Samples were extracted and derivatized in triplicate. Liquid-Liquid Extraction/Cleanup. When the salinity of a sample, as determined by conductivity, might reduce extraction efficiency, an initial liquid-liquid extraction was performed. Aqueous samples were gradually warmed to room temperature and 200 mL was transferred to a 500-mL separatory funnel. The sealed funnel was cooled to 5 °C, acidified with 20 mL of cold sulfuric acid, and then recooled to 5 °C before addition of 100 mL of diethyl ether and saturation with granular sodium sulfate. The sample was then gently agitated for ∼30 s. To a second 500mL separatory funnel, 100 mL of reagent water and 20 mL of 1 M sodium hydroxide were added. Once the layers separated in the first funnel, the lower aqueous phase and all remaining salts were added to a third funnel with the remaining ether phase being decanted into the funnel containing the alkaline reagent water. The initial aqueous sample was extracted in this manner a total of three times with each of the ether layers being transferred to the separatory funnel with the alkaline water. Following agitation and separation of the layers, the alkaline water was drained from the second funnel, along with 2-3 mL of ether, into a sealable container. This final solution was then passed though an anionexchange Empore disk as described above. Following addition to the extraction assembly, the walls of the sample container were rinsed with 3 × 5 mL of reagent water and added to the reservoir. Liquid-liquid extraction of blanks and spikes of reagent water and environmental samples were done a minimum of three times. Samples with a high dissolved organic carbon content can partially plug the disk and lower recoveries. The liquid-liquid cleanup also remedies this problem and is thus recommended for colored samples. Instrumental Conditions. Samples were analyzed on a Perkin-Elmer HS40 headspace sampler coupled to a Perkin-Elmer autosystem GC with an electron capture detector (ECD).12 The data acquisition software used to process and store results was PE Nelson 1020GC Plus. The transfer line connecting the headspace sampler to the analytical column was 2 m of 0.32-mmi.d. deactivated fused silica. The analytical column was a 27 m × 0.32 mm i.d. Poraplot Q (Chrompack, Raritan, NJ) with a 10-µm stationary phase. An additional 1-m segment of 0.32-mm-i.d. deactivated fused silica joined the analytical column to the detector. All capillary column segments were attached by glass joint connectors (Hewlett-Packard, Sunnyvale, CA). The needle and transfer line temperatures on the headspace unit were held constant at 170 °C. Headspace vials were thermostated for 1 h at 50 °C before injection. A column flow rate of 3.2 mL/min (ultrahigh-purity nitrogen) and injection time of 0.1 min resulted in 320 µL of headspace being injected onto the column. The pressurization time for the vial and withdraw time of the needle were 0.5 and 0.2 min, respectively. The GC temperature program used an initial temperature of 110 °C for 10 min followed by a ramp to 210 °C at a rate of 20 °C/min. An additional equilibration time of 15 min after the temperature ramp was sufficient to settle and stabilize the baseline for optimal sensitivity. Finally, the ECD was maintained at 350 °C with 30 mL/min ultrahigh-purity nitrogen makeup gas flowing through. GC/MS Confirmation. A Hewlett-Packard 5970A GC/MS in the selected ion monitoring mode (SIM) was used at 70 eV in 4076 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
the electron impact mode to confirm the presence of TFA. Confirmed samples included Pyramid Lake Marina, NV, Mono Lake, CA, Stillwater National Wildlife Refuge, NV, Lost Lake, CA, and a Selma, CA, rainwater sample. Fragment 69 m/z was scanned in the SIM mode for both standards and environmental samples. The injector port and source temperatures were 170 and 220 °C, respectively. The oven program started at 100 °C and was held for 7 min before being raised to 200 °C at a rate of 20 °C/min where it remained for 2 min more. The analytical column was a 10 m × 0.32 mm i.d. PoraPlot Q (Chrompack) with a constant flow rate of 0.89 mL/min (ultrahigh-purity helium). All of the above samples, with the exception of Reno, NV, rain, were taken though the liquid-liquid extraction and anionexchange Empore disk procedures. The rain was directly extracted with the disk because of its low conductivity. Once sealed in the headspace vial all samples were equilibrated at 50 °C for 1 h. Exactly 1 mL of headspace was manually injected into the instrument using a 1-mL gastight syringe (Hamilton Co., Reno, NV). Injection was performed as quickly as possible to minimize heat loss. A full scan of pure MTFA standard, derivatized TFA standard, and the Lost Lake sample from 55 to 130 m/z was used for additional TFA confirmation. Contamination from Fluorinated Products. PFA (Savillex Corp., Minnetonka, MN) and PTFE fluorinated polymers were tested for their ability to leach TFA into water during prolonged exposure. PTA tubing (19 g/bottle), PTFE lid liners (5 g/bottle), and pieces of PTA used in air sampling (80 g/bottle) were thoroughly cleaned with both deionized water and reagent water before use. Each of these materials was submerged in 500-mL of reagent water in separate amber bottles. Bottle lids were double lined with aluminum foil, capped and stored at room temperature (21 °C) for 2 months. Duplicates of blanks and each of the PFA and PTFE materials were examined. RESULTS AND DISCUSSION The method developed by Zehavi and Seiber12 was modified extensively to reduce manual sample preparation time. This method involved the use of a rotary evaporator to concentrate aqueous samples to near dryness to improve analytical detectablilty for TFA and liquid-liquid back extraction into pure water to remove interferences found in all environmental samples. Dimethyl sulfate (DMS), a carcinogen, was used to methylate TFA to the methyl ester. The evaporation-concentration step was the limiting factor for sample throughput. This process takes 45-120 min for a 250mL sample and requires continual operator involvement to avoid bumping and/or to maximize evaporation. Haloacetic acids such as mono- and dibromoacetic acid and mono-, di-, and trichloroacetic acids have been successfully extracted from water using an anion-exchange resin in EPA method 552.1,15 requiring only a fraction of the time needed for evaporation-concentration. A commercial anion-exchange disk was substituted for resins to simplify conditioning and sample preparation. Elution of the haloacetic acids from the anion-exchange matrix in EPA method 552.1 uses 10% sulfuric acid in methanol. Quantitative methylation was achieved by heating in methyl tert(15) Hodgeson, J. W.; Becker, D. Method 552.1, US EPA, Cincinnati, OH 45269, August 1992.
Table 1. Derivatization Yield and RSD Values for TFA in 4 mL of 10% Sulfuric Acid in Methanola TFA conc (ng/mL)
derivtzn (%)
RSD (%)
TFA conc (ng/mL)
derivtzn (%)
RSD (%)
1.8 8.1 20.3
102.7 108.0 118.5
1.5 1.8 0.4
40.6 80.8 202.8
121.3 110.0 107.0
1.9 0.9 0.3
a
N ) 3.
butyl ether (MTBE). The derivatized acids were then partitioned into MTBE and then analyzed by capillary GC-ECD. Although this technique should work for TFA, the higher volatility of the methyl ester prevents its analysis under the same GC conditions. By coupling this derivatization method with the headspace sampling technique developed by Zehavi and Seiber,12 the TFA methyl ester can be volatilized from the derivatization mixture and analyzed. In addition, replacement of the carcinogenic DMS as a methylating reagent eliminates the high-volatility breakdown products of DMS that can interfere with the automated headspace (HS)-GC determination.12 Derivatization. The capability of a 10% sulfuric acid in methanol solution to derivatize TFA was tested. Quantitative derivatization of TFA to MTFA was obtained for six standards from 1.8 to 202.8 ng/mL, with relative standard deviation (RSD) values for triplicate runs ranging from 0.3 to 1.9% (Table 1). Chromatograms had no interfering peaks, and blanks showed no coeluting compounds (Figure 1). Methanol and air artifacts eluted at least 5 min before the MTFA peak and did not interfere with the analysis. In addition, blanks of methanol in sulfuric acid were indistinguishable from baseline in the retention region of MTFA. Increasing the thermostat temperature to 60 °C apparently increased the quantity of MTFA in the headspace but also increased the quantity of methanol, so that the methanol peak was too large to allow the baseline to stabilize before the elution of MTFA. The temperature of 50 °C was chosen for sample analysis as it provided excellent sensitivity of MTFA and no overlap with the methanol peak. Anion-Exchange Disk Extraction. Extraction of an anionexchange disk within a headspace vial with simultaneous derivatization was examined. A 4-mL sample of 10% sulfuric acidmethanol completely submerged a single disk cut into small pieces. A series of disks were used to extract five 400-mL spike/ standard solutions in reagent water ranging in concentration from 59 to 2345 ng/L. These derivatized disk standards were compared to MTFA and TFA standards subjected to identical derivatization conditions. Quantitative recovery and derivatization was achieved, with RSD values of 0.7-4.4% (Table 2). No additional peaks or significant interferences were observed from the anion-exchange disk. Blanks of the disks conditioned and processed with reagent water produced a small peak averaging 12.8 ng (9.9% RSD; N ) 10) TFA equivalents that eluted at the same time as MTFA. This interference was consistent from batch to batch and over a time period of 2 weeks. A new anion-exchange disk derivatized directly without conditioning and reagent water produced a MTFA peak that was consistent (5.1% RSD; N ) 6) and of a magnitude (11.0 ng) almost identical to the reagent water blank. Thus, no TFA
Figure 1. Representative chromatograms of (a) derivatization blank and (b) derivatized TFA standard. Table 2. Derivatization Yield and Recovery of TFA from 400 mL of Fortified Reagent Watera TFA in 400 mL of reagent water (ng/L)
derivtzn/rec (%)
RSD (%)
234.5 562.8 1173 2345
103.5 115.8 103.5 106.4
1.8 4.4 1.6 0.7
aN
) 3 for all trails.
was picked up through the conditioning process, sample extraction, or drying steps of this procedure. The quantity of TFA determined from the disk blanks was subtracted from the reagent water spike/standard values. We next examined the potential for certain ions such as chloride and sulfate to compete with TFA anion for active sites on the anion-exchange disk.15 Increasing volumes of sodium chloride (400 mg/L) and sodium sulfate (500 mg/L) were tested. Quantitative recoveries were observed for TFA from both spiked salt solutions, with a slight drop-off in recovery for sodium chloride spikes at 300 and 400 mL (Table 3). Two additional 400-mL salt solutions of 1 and 2 g/L for both sodium chloride and sodium sulfate were investigated. The results showed a more pronounced reduction in TFA recovery for the sodium chloride solutions (46 and 12%) and lower recovery for the sodium sulfate solutions (84 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4077
Table 3. Recovery of TFA from High Ionic Strength Solutionsa TFA spiked (ng)
recovery, % (RSD, %) 100 mL
200 mL
300 mL
400 mL
37 463
400 mg/L Sodium Chloride Solutions 99.3 (4.1) 100.3 (6.5) 74.7 (8.5) 104.8 (2.5) 106.2 (4.0) 84.1 (1.8)
61.6 (7.3) 68.8 (4.8)
37 463
500 mg/L Sodium Sulfate Solutions 96.2 (3.5) 109.8 (5.9) 100.6 (1.9) 100.8 (2.8) 100.2 (2.9) 96.9 (0.2)
97.2 (1.9) 86.0 (2.0)
aN
) 3.
Table 4. Recovery of TFA Fortified Environmental Samples and Corresponding Environmental Concentrations Prior to Fortification
sample type chlorinated tap water, Reno, NV rainwater, Selma, CA (combined) 1/96-2/96 fogwater, Selma, CA, 2/4/95 Davis Lake, CA, 8/3/97 Truckee River, NV, at Derby Dam, 9/12/97 Lake Tahoe, CA, at Sand Harbor State Park, 7/25/97 Pyramid Lake, NV, north side, 9/12/97 a
sample fortif environ vol conc reca RSD concb (mL) (ng/L) (%) (%) (ng/L) 200 400 200 400 200 200 400 200 400 200 400 200 400
1157 578.5 1157 578.5 1157 1157 578.5 1157 578.5 1157 578.5 1157 578.5
104.8 1.9 104.4 2.8 100 1.8 94.2 4.2 97.9 1.4 106.3 0.7 102.1 1.3 105.2 1.6 101.9 2.2 102.5 2 94.0 1.5 0 nac 0 na
14.9 14.5 225.7 215.3 2155 86.1 80.9 50.8 55.4 63.1 58.7 na na
N ) 2. b Corrected for disk blank. c na, not applicable.
and 69%), when compared to recoveries of TFA from reagent water with no salt added. Next, the effectiveness of the anion-exchange disk for recovering TFA from spiked, recently collected (1997) environmental samples was examined. A single spike of 231.4 ng of TFA at 200and 400-mL volumes was performed. Prior to disk extraction, all samples were filtered with a GF/F-type filter to remove particulate matter. The results from chlorinated tap water, rainwater, fogwater, and several surface waters are in Table 4. Quantitative recovery was achieved with all spiked samples except that from Pyramid Lake, NV, which gave an even lower response than reagent water or disk blanks. This is attributed to the high concentration of salt (conductivity 10 380 µS, >5 g/L NaCl equivalents) present in this sample and the complete elimination of TFA due to anion competition and saturation of the anionexchange disk. This high ionic strength solution also eliminated the TFA that was found with disk blanks. Samples such as Pyramid Lake must undergo the liquid-liquid extraction before passing through the disk. Liquid-Liquid Extraction. To determine whether an additional cleanup by liquid-liquid extraction is necessary, conductivity measurements were compared to a set of sodium chloride standards in water. Recovery of TFA began to diminish with 120 mg of sodium chloride in 400 mL of water passing through the disk. A standard curve of conductivity measurements was used to determine the salinity content at which liquid-liquid extraction 4078 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
Figure 2. Chromatogram of Pyramid Lake, NV, sample processed by liquid-liquid back extraction and the anion-exchange disk.
was needed. Environmental samples above 620 µS, corresponding to a concentration of 300 mg/L sodium chloride, were liquidliquid extracted before passing through the anion-exchange disk. As a conservative measure, all samples with conductivity readings over 500 µS were processed in this manner. Liquid-liquid extraction was used with samples taken from Pyramid Lake, NV, Mono Lake, CA, and the Pacific Ocean at San Francisco, CA, prior to processing by the anion-exchange disk. The chromatogram for one of the Pyramid Lake trials is shown in Figure 2. The results (Table 5) showed excellent recoveries and reproducibility for all three samples. Mono Lake (pH 10.0, >120 000 µS) had lower recovery apparently because of the vigorous release of carbon dioxide upon addition of the sulfuric acid, resulting in some loss of TFA. Limit of Detection and Quantification. The limits of detection (LOD) and quantification (LOQ) were determined by the extraction of a series of spiked reagent water samples. The LOD was determined by adding the y-intercept area from the standard curve to 3 times the standard deviation of three blanks. The 11-ng (sd ) 0.56) disk blank represents the actual background of a water sample processed by the disk. Therefore, the LOD (blank + 3 sd) is 12.7 ng and the LOQ (blank + 6 sd) is 14.4 ng. For a 400-mL sample, the LOD and LOQ are 32 and 36 ng/L, respectively. Consistency. Since Mono Lake was one of the most difficult samples to process, it was used to validate the overall consistency of the technique. The RSD for 10 samples processed by liquidliquid extraction, anion-exchange disk extraction, and then headspace analysis was 6.9%. Since the inherent concentration of TFA in the Mono Lake sample was relatively high (198 ng/L) and falls within the second third of the standard curve, the results were expected to be relatively consistent. A small change in detector response within the upper region of the standard curve will correspond to a small change in concentration. At the low end of the standard curve, this small change will be more pronounced. A single sample taken from the Truckee River in Verdi, NV, was passed directly through the anion-exchange disk and found to contain very low concentrations of TFA (29.5 ng/L). At this level, the nine replicate analyses produced a greater RSD value (13.7%), as expected.
Table 5. Liquid-Liquid Extraction Spike Recovery and Consistency sample type
sample vol (mL)
fortif conc (ng/L)
reca (%)
RSD (%)
environ conc (ng/L)
RSD (%)
reagent water Pyramid Lake, NV, at Marina, 9/12/97 Pyramid Lake, NV, south end, 9/12/97 Mono Lake, CA, 9/27/97 Pacific Ocean, San Francisco, CA, 2/10/97
200 200 200 200 200
578.5 578.5 none 578.5 681.5
100.8 95.7 na 89.0 103.8
1.8 6.2 na 1.6 5.9
nab 81.3 (N ) 5) 86.5 (N ) 2) 198.0 (N ) 10) 275.0 (N ) 4)
na 6.3 14.1 6.9 7.9
a
N ) 3. b na, not applicable.
Table 6. Comparison of TFA Concentrations with Samples Analyzed by the Zehavi and Seiber Technique sample
analysis method
conc (ng/L)
diff (%)
Selma, CA, fogwater collected 2/4/95 Woodward Lake, CA, collected 1/4/96
Zehavi and Seiber Wujcik and Seiber Zehavi and Seiber Wujcik and Seiber
1640 (N ) 1) 2154 (N ) 2) 576 (N ) 1) 550 (N ) 2)
24a
a
4.5
Possible contamination from previous operations.
Table 7. Comparison of GC/MS SIM Results to HS-GC-ECD
sample Reno, NV, rain, 6/13/97 start 6:00 p.m. Pyramid Lake, NV, 9/12/97 Stillwater National Wildlife Refuge, NV, 9/16/97 Mono Lake, CA, 9/27/97 Lost Lake, Fresno County, CA, 12/27/96a a
conc (ng/L) diff GC/MS HS-GC-ECD (%) 196 96 429
149 87 426
152 13071
198 9877
24 9.4 0.7 23 24
Not replicated.
Two samples from Woodward Lake and a single fogwater sample from Selma, CA, were compared to previous results determined by Wujcik et al.14 with the analytical and extraction method developed by Zehavi and Seiber12 (Table 6). Concentrations of TFA in the Woodward Lake samples were very similar with only 4.5% difference between the average of the anionexchange disk extraction technique and the evaporationconcentration method. The slightly higher concentration (24% difference) in the fogwater sample can be attributed to exposure to airborne TFA during opening and filtering operations at an earlier date.16 Confirmation of TFA. Environmental samples of TFA were also quantified using GC/MS in the SIM mode scanning for the 69 m/z (trifluoromethyl) fragment. Other MS fragments were either not abundant enough or raised the background signal, preventing effective quantification at lower TFA concentrations. GC/MS concentrations were similar to the HS-GC-ECD results (Table 7). Differences in concentrations are a result of the error involved in manual injection onto the GC/MS and the cooling of samples taken from the oven during the injection process. Since it was necessary to heat the samples to drive enough TFA to the headspace to quantify the lower concentrations, the error from (16) Zehavi, D.; Seiber, unpublished work, 1996.
cooling could not be avoided when samples were injected manually. Overall, quantification of the environmental samples by GC/MS agreed with results of the HS-GC, confirming that TFA is present and no coeluting compounds interfere. Scans from 55 to 130 m/z for a higher concentration sample obtained from Lost Lake, CA, matched well with standards of MTFA and derivatized TFA. A full scan was not possible with the other samples due to their lower TFA concentrations. This additional confirmation shows that the peak that results from the derivatization of TFA is MTFA, which is also the peak present in the derivatized environmental samples. Sample Throughput. Triplicate 400-mL samples extracted with the anion-exchange disk side by side can be processed in ∼20 min, including cleaning of the extraction apparatus. This was a marked improvement over the previous evaporation-concentration technique,12 which required 45-120 min per 250-mL sample. Furthermore, samples that were above the conductivity/salinity limit (500 µS) could be processed in triplicate in ∼60 min including liquid-liquid extractions. Samples such as Mono Lake may require slightly more time due to the evolution of carbon dioxide. Contamination from Fluorinated Materials. Each of the precleaned PFA and PTFE materials tested showed an increase in TFA over blanks. PTFE caps elevated the level of TFA by 8.0 ng for each gram present. Acid-washed PFA tubing yielded 12.6 ng of TFA for each gram of tubing, and a single trial of PFA air sampling connectors produced 0.6 ng of TFA for each gram. If TFA is being leached from these materials, their mass, surface area, thickness, and density will affect the rate of flux. Since PFA and PTFE materials may leach TFA with continual water contact, it is important to avoid their use in the sampling and storage of aqueous media. Field blanks using aluminum foil lining between the glass bottle and cap showed no increase in TFA over a 1-year period. This is the recommended method for long-term storage. CONCLUSIONS This improved extraction and analytical technique can be used to quantify TFA in all aqueous environmental samples tested. The low detection limit, high sample throughput (3 samples/20 min), and ease of use make the technique practical for routine sample analysis. Conductivity measurements eliminate guesswork on the need for the liquid-liquid extraction procedure. In addition, the method uses relatively hazard-free reagents. Environmental samples processed and analyzed by this technique show relatively low levels of TFA (50-86 ng/L) in more pristine waters such as Lake Tahoe and in dynamic systems such as Davis Lake and the Truckee River. Pyramid Lake, the terminal end point of the Truckee River and its tributaries, accumulates Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
4079
salts from input and seasonal evaporation.9 A single sample from Pyramid Lake collected on 7/14/94, reported by Zehavi and Seiber,12 showed an extremely high concentration of 41 µg/L TFA. Replicate samples from two locations on the lake (9/12/97) in this study gave only 81.3 and 86.5 ng/L. We do not know why such a difference in concentration for samples taken only 2 years apart existed. Mono Lake had a higher level of TFA (198 ng/L) than the 1997 Pyramid Lake samples. Mono Lake has a higher evaporation rate and lower volume than Pyramid Lake, apparently concentrating TFA to a greater extent. Pyramid Lake and Mono Lake should provide information on the potential for lake systems with no output to accumulate TFA.
4080 Analytical Chemistry, Vol. 70, No. 19, October 1, 1998
ACKNOWLEDGMENT Funding for this research was provided in part by an NSF Grant (ATM-9708504) and a NSF EPSCoR fellowship (1330-15352AT) awarded to C.E.W. The authors also thank the Sierra Pacific Power Co. for general support of the Center for Environmental Sciences and Engineering.
Received for review February 23, 1998. Accepted June 30, 1998. AC9802123
