Environ. Sci. Technol. 2000, 34, 3115-3118
Natural Background Levels of Trifluoroacetate in Rain and Snow LENA M. VON SYDOW,* A N D E R S B . G R I M V A L L , H A N S B . B O R EÄ N , KRZYSZTOF LANIEWSKI, AND ANNIKA T. NIELSEN Department of Water and Environmental Studies, Linko¨ping University, SE-581 83 Linko¨ping, Sweden
It has been shown that some of the fluorinated ethane derivatives being introduced as CFC-replacements can be transformed to TFA (trifluoroacetate) in the atmosphere. Moreover, TFA is extremely stable in the environment, and this has raised questions regarding how widespread TFA is in the environment. We found that TFA is ubiquitous in precipitation: samples of rain from Ireland and Poland and snow from Canada, Sweden, New Zealand, and East Antarctica contained 1-1100 ng/L, and, studying a firn core drilled in Antarctica, concentrations of 3-56 ng/L were measured in layers formed during the 19th century. We have confirmed the preindustrial presence of significant background concentrations of trifluoroacetate in historic precipitation samples from the analysis of firn. Extensive procedures were enforced to prevent sample contamination.
Introduction Some of the halogenated ethane derivatives introduced as environmentally friendly alternatives to CFCs (chlorofluorocarbons) may be responsible for the formation of persistent degradation products. In particular, it is assumed that HCFC123 (CCl2HCF3), HCFC-124 (CClHFCF3), and HFC-134a (CF3CFH2) are transformed into trifluoroacetylfluoride and subsequently hydrolyzed to TFA (trifluoroacetate) (1-4 ). Assessments of possible pathways of TFA indicate that hydrolysis, photolysis, and formation of insoluble salts play a minor role (5a,b,6). Microbial degradation might be a more significant process according to Visscher et al. (7), but later studies have failed to repeat these results (8). Hence, it is possible that a large-scale introduction of HCFCs will eventually lead to accumulation of TFA in lakes and wetlands with little outflow and high evaporation (4,9). Therefore, it is important to determine both the current concentrations and the background levels of TFA in precipitation. Frank and co-workers (10,11) reported that TFA levels in rainwater collected in Germany and Switzerland ranged from < 3 to 280 ng/L. In the United States, Zehavi and Seiber (12) found TFA concentrations of 31-280 ng/L in fog and rainwater in Nevada and California, and, more recently, Wujcik et al. (13) detected between 31 and 3779 ng/L in fog, rain, and snow collected in the same two states. Additionally, Scott and Alaee (14) found up to 520 ng/L in samples of rain and snow from Ontario and the Northwest Territories in Canada. Even more recently, Jordan and Frank (15) reported TFA concentrations between 10 and 410 ng/L in precipitation collected in Germany. * Corresponding author phone: +46-13-28 10 00; fax: +46-13-13 36 30; e-mail:
[email protected]. 10.1021/es9913683 CCC: $19.00 Published on Web 06/29/2000
2000 American Chemical Society
TABLE 1. Physical Characteristics of Firn Sampled in Dronning Maud Land, East Antarctica (74°59′59′′S, 15°00′06′′E)a sample
mean depth (m)
density (kg/L)
volume (L we)
M2 M4 M 11 M 15 M 22 M 31 M 35 M 36 M 39 M 42 M 45
2.6 3.9 6.9 8.4 10.9 13.9 15.1 15.8 16.9 17.7 18.8
0.369 0.375 0.438 0.464 0.486 0.507 0.558 0.532 0.544 0.540 0.557
0.263 0.394 0.690 0.343 0.223 0.268 0.238 0.379 0.320 0.386 0.377
a The volume of the firn samples is expressed in water equivalents (we).
The objectives of the present study were (i) to investigate the presence of TFA in precipitation collected at remote sites and (ii) to test the hypothesis that natural sources play a significant role in the current background levels. With this end in view, gas chromatography with mass spectrometric detection (GC-MS) was employed to determine TFA levels in rain, snow, and glacier ice and in different segments of a 20-m firn core drilled in East Antarctica.
Materials and Methods Sampling. Samples of snow and rain were collected in several remote areas, including East Antarctica, northern Canada, northern Sweden, and western Ireland as well as in central Poland. Snow was sampled by pressing 2-L polyethylene bottles horizontally through snowpacks just below the surface layer. Rainwater was collected using a funnel-shaped stainless steel sampler (surface area 3 m2) that was emptied within 24 h after each rain event. Glacier ice was taken from the subpolar Mårmaglacia¨ren in northern Sweden. This was achieved by removing about 1.5 m of snow and 20 cm of the ice that was closest to the surface of the glacier and then using a chain saw to cut out large blocks of solid ice (5-27 L). Samples of firn were derived from 40 cm lengths of a 20-m core that was drilled during the 1996-1997 Norwegian Antarctic Research Expedition (NARE) organized within the framework of the European Project for Ice Coring in Antarctica (EPICA), see Table 1. The 10- to 40-cm-long core segments were packed in polyethylene bags (0.15 mm; Labora, Stockholm). Three of the segments (M 31, M 36, and M 45) were further protected from contamination during transport and storage by putting the plastic bags in stainless steel cylinders made airtight with a copper gasket (Leybold Inc., Gothenburg, Sweden). All samples were transported and stored frozen (-20 °C) pending analysis. To reduce the risk of sample contamination, a handsaw was used to remove the outer 5 cm of the glacier ice prior to chemical analysis. Likewise, a 50-mm inner core was acquired from the original 75-mm firn core by using a specially designed device made of Teflon and stainless steel (16). Dating of Firn and Glacier Ice. The age of the sampled firn was calculated from measurements made on a core drilled next to the core analyzed in the present study. Information about the depth of the layer dating from 1954 to 1955 was obtained from records of β-activity, and electrical conductivity measurements (ECM) were used to reveal firn layers VOL. 34, NO. 15, 2000 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3115
FIGURE 1. GC-MS-SIM chromatograms illustrating responses to the molecular ion (m/z 308) of the 1-(pentafluorophenyl)ethyl ester of TFA in extracts derived from firn collected in Dronning Maud Land, Antarctica (left tracing), and from Milli-Q water used as a blank (right tracing). The age of the firn sample was approximately 170 years, and the concentration of TFA was 38 ng/L. formed after major volcanic eruptions. The deepest part of the firn core (20.2 m) was estimated to represent snow that accumulated about 190 years ago. Glacier ice was sampled where clearly visible annual growth layers enabled identification of sections of ice that had been formed at least 500 years ago (17).
Analytical Procedures The analytical procedure employed to determine TFA in precipitation was a modification of the method described by Frank and co-workers (10) and comprised the following steps: rotary evaporation of the original sample to dryness, addition of toluene, derivatization of the extracted trifluoroacetic acid, and GC/MS detection of the 1-(pentafluorophenyl)ethyl ester of TFA. The derivatizing agent, 1-(2,3,4,5,6pentafluorophenyl)diazoethane, was prepared and utilized according to Meese (18) and Hofmann and co-workers (19). The volumes of the analyzed samples of rain and melted snow and ice varied from 200 to 1000 mL. The individual samples were treated as follows: an internal standard (2,2dichloropropionic acid, 500 ng) was added, and the sample was placed in a pear-shaped flask and evaporated in a rotary evaporator; to avoid loss of TFA to the air during the final phase of the evaporation, the concentrated sample was transferred to a smaller flask (25 mL), and a phosphate buffer (pH 5.9) was added to control the pH of the sample. Toluene (100 µL) and the derivatizing agent 1-(2,3,4,5,6-pentafluorophenyl)diazoethane (0.3 M in benzene) were then added, and the extract was subjected to ultrasonification for 15 s. After 2 h at room temperature, hexane (100 µL), Milli-Q water (100 µL), and a second internal standard (500 ng of 1-chlorooctane, to check the GC analysis) were added. The final hexane extract was analyzed on a Hewlett-Packard 6890 GC equipped with an HP 5973 mass spectrometer in the selected ion monitoring (SIM) mode. The following ions were monitored: 69, 195, 308 (TFA); 97, 99, and 101 (2,2dichloropropionic acid); and 91, 93, and 105 (1-chlorooctane). Gas chromatographic conditions: HP-5 column (30 m * 0.25 mm, phase thickness 0.25 µm); carrier gas helium, flow rate 39 cm/s; pulsed splitless injection (1 min), 1 µL injected manually; injector temperature 250 °C; temperature program 30 °C (5 min), 5 °C/min, 105 °C; transfer line temperature 280 °C. Quantification of the TFA in analyzed samples was based on the area ratio of ions 308 and 97. Calibration functions were established by analyzing a total of 12 10-mL samples of Milli-Q water spiked with 0, 100, 200, or 300 ng of TFA. The response was found to be almost linear in the indicated concentration range, and the R 2 value was 0.96. Blank levels were always < 1 ng/L of TFA, and the limit of detection, defined as the average blank level plus three standard deviations of the blank, corresponded to 1 ng. 3116
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 34, NO. 15, 2000
FIGURE 2. Measured concentrations of TFA in Antarctic firn collected at different depths. Sample characteristics are given in Table 2.
Results GC/MS/SIM analysis allowed detection of the 1-(pentafluorophenyl)ethyl ester of TFA at concentrations as low as 1 ng/L. The SIM chromatograms in Figure 1 illustrate the responses to the molecular ion (m/z 308) shown by a sample of firn from Dronning Maud Land, Antarctica, and a blank sample of Milli-Q water. Further analyses showed that TFA was present in all of the firn samples from Antarctica; concentrations ranged from 6 to 56 ng/L, with an average of 25 ng/L. As can be seen in Figure 2, there was no obvious relationship between concentration and depth in the firn core. The snow samples collected in the Northern and Southern Hemispheres varied markedly in regard to TFA levels, see Table 2. The mean concentration of TFA in snow from Sweden and Canada was 10 ng/L, and the corresponding value for Antarctica was 40 ng/L. In addition, TFA was detected at a concentration of 5 ng/L in glacier ice from a semipolar glacier in northern Sweden. The concentrations of TFA in individual rainfalls varied even more markedly, see Table 3. At Mace Head, on the west coast of Ireland, observed levels ranged from 2 to 92 ng/L of TFA. Rain collected in the city of Gdansk, Poland, contained up to 1100 ng/L of TFA, thus indicating the presence of a local source.
Discussion The results of the GC-MS analyses show that the analytical procedure employed was adequate to determine TFA in precipitation collected at remote sites. The response to the ions used for TFA detection (m/z 69, 195, and 308) was invariably larger for real samples than for any of the accompanying Milli-Q blanks, and the peaks used for quantification were well resolved and separated from other peaks in the retention intervals under consideration (see Figure 1). Furthermore, the low and stable blank level (