Anal. Chem. 1999, 71, 4465-4471
Simplified Method for Trace Analysis of Trifluoroacetic Acid in Plant, Soil, and Water Samples Using Headspace Gas Chromatography Thomas M. Cahill, Jody A. Benesch, Mae S. Gustin, Erica J. Zimmerman, and James N. Seiber*,1
Center for Environmental Sciences and Engineering/Mailstop 199, University of Nevada, Reno, Nevada 89557
A simple and sensitive analytical procedure was developed to determine the concentration of trifluoroacetic acid (TFA) in plant, soil, and water samples. The analysis involves extraction of TFA by sulfuric acid and methanol followed by derivatization to the methyl ester of TFA (MTFA). This is accomplished within a single vial without complex extraction procedures. The highly volatile MTFA is then analyzed using headspace gas chromatography. The spike recovery trials from all media ranged from a low of 86.7% to a high of 121.4%. The relative standard deviations (RSDs) were typically below 10%, although some tests had higher RSDs due to low concentrations near the detectable limit. The minimum detectable limit (MDL) for the method was 34 ng/g for dry plant material, 0.20 ng/g for soil, and 6.5 ng/L for water. This is the first practical method for low-level analysis of TFA in plants and soil, and it is a significant improvement over previous methods for TFA in water. In the early 1990s, chlorofluorocarbon (CFC) refrigerants and propellants were largely replaced by hydrofluorocarbons (HFCs) and hydrochlorofluorocarbons (HCFCs) in order to reduce stratospheric ozone depletion.1 The HFCs and HCFCs, unlike the older CFCs, are unstable in the troposphere and can degrade to form trifluoroacetic acid (TFA, CF3COOH) as a byproduct.2 As a result of its high water solubility and low Henry’s constant, trifluoroacetic acid is removed from the atmosphere primarily through wet deposition3,4 where it tends to accumulate in aquatic ecosystems with little outflow or seepage and high evaporation rates.5,6 TFA has two characteristics that are cause for environmental concern: exceptional stability and phytotoxicity. TFA’s stability arises from its effective immunity to environmental oxidation and * Corresponding Author. Fax: (775) 784-1142. E-mail:
[email protected]. 1 Present address: USDA-ARS, 800 Buchanan St., Albany, CA 94710-1105. (1) Montreal Protocol on Substances that Deplete the Ozone Layer, 16 Sept. 1987, Reprinted In: Weiss, E. B., Ed. International Environmental Law: Basic Instruments and References; Transnational Publishers, Inc.: U.S.A. (2) Wallington, T. J.; Schneider, W. F.; Worsnop, D. R.; Nielsen, O. J.; Sehested, J.; Debruyn, W. J.; Shorter, J. A. Environ. Sci. Technol. 1994, 28, 320A325A. (3) Bowden, D. J.; Clegg, S. L.; Brimblecombe, P. Chemosphere 1996, 32, 405420. (4) Kotamarthi, V. R.; Rodriguez, J. M.; Ko, M. K. W.; Tromp, T. K.; Sze, N. D.; Prather, M. J. J. Geophys. Res. 1998, 103, 5747-5758. (5) Tromp, T. K.; Ko, M. K. W.; Rodriguez, J. M.; Sze, N. D. Nature (London) 1995, 376, 327-330. (6) Schwarzbach, S. E. Nature (London) 1995, 376, 297-298. 10.1021/ac990484l CCC: $18.00 Published on Web 09/18/1999
© 1999 American Chemical Society
reductive dehalogenation.7,8 As a result of its stability, it has been demonstrated that TFA accumulates in terminal water bodies.5,9,10 Currently, typical TFA concentrations in surface water range from approximately 100 to 500 ng/L, although TFA concentrations as great as 6.4 µg/L have been found in certain terminal lakes.9,11 TFA has been shown to impact sensitive species of algae, such as Raphidocelis subcapitata, with a lowest observed effective dose of 360 µg/L.8 Most plant species tested were impacted by TFA at concentrations in the mg/L range.8,12 Although these toxic concentrations are 2-3 orders of magnitude higher than typical current environmental levels,9-11 there is concern that TFA may accumulate over many years in terminal water bodies to the point that concentrations toxic to algae and some higher plants may be achieved.5,6 Tromp et al.5 predict that TFA concentrations in seasonal wetlands with little or no outflow may exceed 100 µg/L after 30 years, thus approaching toxic levels for sensitive species. Since plants have been shown to readily take up and accumulate TFA,13 a simple analysis method for quantifying TFA within plants at environmental concentrations is needed. Determination of TFA concentrations in environmental soil and water samples is also needed because these are pathways of exposure to vegetation as well as environmental compartments that tend to retain TFA.14,15 TFA has been analyzed in various media using ion chromatography (IC),16-18 capillary electrophoresis (CE),19 19F nuclear (7) Emptage, M.; Tabinowski, J.; Odom, J. M. Environ. Sci. Technol. 1997, 31, 732-734. (8) Boutonnet, J. C.; Bingham, P.; Calamari, D.; de Rooij, C.; Franklin, J.; Kawano, T.; Libre, J.; McCulloch, A.; Malinverno, G.; Odom, J. M.; Rusch, G. M.; Smythe, K.; Sobolev, I.; Thompson, R.; Tiedje, J. M. Hum. Ecol. Risk Assess. 1999, 5, 59-124. (9) Jordan, A.; Frank, H. Environ. Sci. Technol. 1999, 33, 522-527. (10) Wujcik, C. E.; Cahill, T. M.; Seiber, J. N. Environ. Sci. Technol. 1999, 33, 1747-1751. (11) Frank, H.; Klein, A.; Renschen, D. Nature (London) 1996, 382, 34. (12) AFEAS(Alternative Fluorocarbons Environmental Acceptability Study), Workshop on the Environmental Fate of Trifluoroacetic Acid, Miami Beach, FL, March 3-4, 1994; Chumley, Forrest, Ed.; AFEAS: Washington, D.C., 1994. (13) Rollins, A.; Barber, J.; Elliott, R.; Wood, B. Plant Physiol. 1989, 91, 12431246. (14) Likens, G. E.; Tartowski, S. L.; Berger, T. W.; Richey, D. G.; Driscoll, C. T.; Frank, H. G.; Klein, A. Proc. Natl. Acad. Sci. U.S.A. 1997, 94, 4499-4503. (15) Berger, T. W.; Tartowski, S. L.; Likens, G. E. Environ. Sci. Technol. 1997, 31, 1916-1921. (16) Hankins, D. C.; Kharasch, E. D. J. Chromatogr., B 1997, 692, 413-418. (17) Kawaguchi, R.; Fujii, K.; Morio, M.; Yuge, O.; Hossain, D. Hiroshima J. Med. Sci. 1989, 38, 27-34. (18) Richey, D. G.; Driscoll, C. T.; Likens, G. E. Environ. Sci. Technol. 1997, 31, 1723-1727.
Analytical Chemistry, Vol. 71, No. 20, October 15, 1999 4465
magnetic resonance (NMR),13,20,21 and gas chromatography (GC).22-26 While IC, CE, and NMR analyses are simple and direct, they lack the sensitivity to measure TFA at current environmental concentrations. GC methods, while very sensitive, often require extensive sample cleanup and derivatizations. The analytical technique for TFA in plant, soil, and water samples described here achieves the simplicity of the IC method and the sensitivity of the GC techniques required to determine TFA in environmental samples. The analysis procedure involves acid digestion, which extracts the TFA from the media, followed by a methylation derivatization and analysis by headspace gas chromatography (HS-GC). The same analysis procedure can be used on plant, soil, and water samples with only minor modifications. Previous HS-GC analysis methods have not been applied to solid samples, such as plants or soil, and the aqueous analysis procedures required extensive operator time,25 expensive reagents,26 or lacked sensitivity necessary for environmental sample analysis.23,24 One haloacetic acid GC analysis procedure has been presented for plants, but it is simply a water wash of ground plant material after which the water is analyzed.27 Considering that TFA is retained by organic matter15,18 and may be incorporated into biomolecules,28 a simple water wash may not be sufficient to dislodge TFA incorporated into the plant matrix. The method reported here is a combined extraction/derivatization conducted within a single vial without the need for filtering, cleanup, or other time-consuming transfer steps. EXPERIMENTAL METHODS Materials. The sulfuric acid (96.5%), anhydrous sodium sulfate (99.7%), ammonium carbonate (HPLC grade), and methanol (99.9%) were obtained from Fisher Scientific (Pittsburgh, PA). Purified sand was purchased from J. T. Baker (Phillipsburg, NJ). Sodium trifluoroacetate (99+%) was obtained from Sigma Chemical Co. (St. Louis, MO). A TFA stock solution was created by dissolving 1.204 g of the sodium salt of TFA in 1 L of methanol, thus yielding 1 g of TFA per liter of methanol. Twelve TFA standards, ranging from 0.125 ng/mL to 2000 ng/mL, were created through dilutions of the stock solution with methanol. Reagent water (18 MΩ‚cm specific resistance) was obtained from a Barnstead Nanopure system (Dubuque, IA) and was further distilled prior to use. This reagent water was free of detectable TFA. All sample and chemical weighing was conducted using a Mettler AE 50 balance, the accuracy of which was tested weekly. Liquid reagents were measured by two Oxford Labware variable pippettors (St. Louis, MO) with 1-1000 and 1-5000 µL ranges (19) Strege, M. A.; Mascher, W. G. J. Chromatogr., B 1997, 697, 255-257. (20) London, R. E.; Gabel, S. A. Biochemistry 1989, 28, 2378-2382. (21) Monte´, S. Y.; Ismail, I.; Mallett, D. N.; Matthews, C.; Tanner, R. J. N. J. Pharm. Biomed. Anal. 1994, 12, 1489-1493. (22) Karashima, D.; Shigematsu, A.; Furukawa, T.; Nagayoshi, T.; Matsumoto, I. J. Chromatogr. 1977, 130, 77-86. (23) Maiorino, R. M.; Gandolfi, A. J.; Sipes, I. G. J. Anal. Toxicol. 1980, 4, 250254. (24) Gruenke, L. D.; Waskell, L. A. Biomed. Environ. Mass Spectrom. 1988, 17, 471-475. (25) Zehavi, D.; Seiber, J. N. Anal. Chem. 1996, 68, 3450-3459. (26) Wujcik, C. E.; Cahill, T. M.; Seiber, J. N. Anal. Chem. 1998, 70, 40744080. (27) Frank, H.; Scholl, H.; Renschen, D.; Rether, B.; Laouedj, A.; Norokorpi, Y. Environ. Sci. Pollut. Res. Int. 1994, 1, 4-14. (28) Standley, L. J.; Bott, T. L. Environ. Sci. Technol. 1998, 32, 469-475.
4466
Analytical Chemistry, Vol. 71, No. 20, October 15, 1999
unless otherwise noted. Sample Collection and Preparation. Five species of plants used to test the accuracy and efficiency of the analytical method were the following: (1) a herbaceous annual (monkeyflower, Mimulus guttalus); (2) an annual grass (beard grass, Polypogon monspeliensis); (3) a deciduous tree (black locust, Robinia pseudoacacia); (4) a coniferous tree (Ponderosa pine, Pinus ponderosa); and (5) a cereal grain (rice, Oryza sativa). The monkeyflower, beard grass, and rice plants were grown hydroponically in 0.25 Hoagland solution.29 Five plants from each species were grown in solutions containing 0, 8.3, 83, and 830 µg/L TFA. The plants were grown for 36-48 days, depending on the species, after which the entire plant was collected, and the roots were removed. The above-ground part of all five plants in each treatment was frozen in liquid nitrogen and homogenized into a single powder. The powder was then lyophilized for 24 h. Ten aliquots of this powder were analyzed in each trial of the analytical technique. The plants grown in solutions with no added TFA were used for determining background TFA concentrations and for spike-recovery tests. The black locust samples were collected during summer from a mature tree grown on the University of Nevada campus. The Ponderosa pine samples were obtained from 11/2 year old trees grown in a greenhouse. The samples from both species were powdered and freeze-dried. Neither of the tree species had controlled TFA dosing, so a single homogenized powder was obtained for each species for spike-recovery tests. The rice samples were divided into two sets to determine if processing the plant samples as wet clippings would differ from processing them as dried and powdered samples. The first set was powdered and dried as usual, while a second set was processed as wet samples. The wet-weight samples were cut up as wet weight into a sample vial, capped, and frozen at -10 °C until analysis. To determine the wet weight/dry weight ratio, 10 rice plants were weighed, freeze-dried for 24 h, and subsequently reweighed to obtain the dry weight. The soil analysis procedure was tested using (1) purified sand; (2) humic forest soil from Tahoe National Forest, CA; (3) alkali playa sand from Stead, NV; and (4) agricultural soil from Nevada Cooperative Extension fields in Reno, NV. The soil samples were homogenized and sifted. Thirty aliquots of each soil type were used for spike-recovery tests. The water samples were (1) distilled water originating from a Barnstead Nanopure water system; (2) rainwater collected between February 6, 1999 and February 27, 1999 in Davis, CA; (3) water collected from Mono Lake, CA; and (4) a vernal pool at the Yolo County Regional Grassland Park, CA. The first two samples represent relatively clean water samples, while the last two samples are terminal water systems. Plant Sample Analysis. Plant samples were processed by adding 0.05 g of dry plant material and 0.4 g of sodium sulfate to a 22-mL HS-GC vial (Perkin-Elmer, Norwalk, CT). Three milliliters of 9 M sulfuric acid were added to the vial. The acid digests the plant material and is part of the derivatization solution. TFA may be incorporated into biomolecules as an amide-bonded trifluoroacetyl moiety,28 thus the acid solution will help to cleave the amide (29) Cramer, G. R.; Bowman, D. C. J. Exp. Bot. 1991, 42, 1417-1426.
Figure 2. Chromatograms from (A) 0.01 g of a monkeyflower grown in 8.3 µg/L TFA hydroponics using the time-optimized GC conditions and (B) 1 g of agricultural soil sample using the sensitivity optimized conditions. The mass of MTFA represented by the peak is displayed on the chromatogram. These masses correspond to a concentration of 36.2µg/g for the plant sample and 1.2 ng/g for the soil sample.
Figure 1. Chemical reactions associated with TFA extraction and derivatization.
bonds and free this bound TFA. Since TFA has been shown to bind to organic matter,18 sodium sulfate was added to help displace the TFA anions from the plant material and help to “salt out” the MTFA from the solution. The vial was then capped and placed on an orbital shaker (Thermolyne, Dubuque, IA) at 300 rpm for 24 h. After the plants had been digested, the samples were frozen at -10 °C for 1 h. This step condenses any protonated TFA that might exist in the vapor phase prior to uncapping the vial. The vials were then uncapped and 1 mL of methanol was added, thus completing the 75% acid/25% methanol derivatization solution. The vials were quickly recapped and shaken again for another 24 h, after which the samples were ready for analysis by HS-GC (Figure 1). The wet-weight samples were processed in the same manner except that 0.5 g of wet-weight plant clippings were added to the vials instead of 0.05 g of dried powder. Reagent blanks, which were processed exactly like samples except that they lacked plant material, showed no detectable TFA in all twelve trials. This demonstrated that the reagents, namely methanol, sodium sulfate, and sulfuric acid, were free of detectable TFA contamination. This digestion procedure cannot break down the structural lignins of the plant, which remain as a heterogeneous mixture in the vial. The suspended plant particles do not adversely affect the analysis so long as all the plant material is soaked and saturated with the acid/methanol solution. TFA is permeable through cell membranes by diffusion,20 thus it can escape into solution even if digestion is incomplete. Fine clipping or grinding of tough samples, such as pine needles, speeds the extraction of TFA from the plant particles.
To conduct spike-recovery tests on the plants, the nondosed plant samples were placed in an HS-GC vial along with 1 mL of methanol spiked with either 25 or 100 ng of TFA. A third set of control plant samples had 1 mL of pure methanol added to them. The 1 mL of spiking solution, taken from TFA standard solutions, was measured using a 1-mL Kimax glass pipet. These spike concentrations correspond to 500 and 2000 ng/g of dry plant material. The plant sample was allowed to soak in the methanol solution for 1 h, thus allowing the TFA to permeate into the plant. The methanol was then gently evaporated at 50 °C for 3 h leaving the dried plant material. Analyses were conducted by a Perkin-Elmer Autosystem gas chromatograph using an HS-40 headspace autosampler and an electron capture detector (ECD), as reported previously.26 But in the present method, the HS-40 headspace autosampler was equipped with a high-pressure sampling accessory that allowed the vial pressure to be set independently of the column-head pressure. The sample vial was thermostated on the autosampler at 50 °C for 1 h prior to analysis. The sample vial was pressurized at 170 kPa for 0.5 min prior to injection of the headspace vapor onto the column. An injection time of six seconds caused 0.33 mL of the headspace vapor to be delivered to the column. Both the needle temperature and transfer line of the headspace autosampler were maintained at 170 °C. The GC column was a 25 m × 0.32 mm Poraplot Q HT, with a 10-µm film thickness, from Chrompack (Raritan, NJ). High-purity nitrogen was used for both the carrier and makeup gas. The carrier flow was 3.3 mL/min, and the makeup flow was 30 mL/ min. The detector temperature was maintained at 350 oC. The oven program was a 10-min isothermal period at 110 °C followed by a 40 °C/min ramp to 250 °C where it was held for a 10-min bakeout period. Under these conditions, MTFA elutes from the column at 8.0 min. (Figure 2A). An additional 10 min between runs was allowed for signal equilibration. This program is optimized for Analytical Chemistry, Vol. 71, No. 20, October 15, 1999
4467
relatively rapid analysis runs. The column slowly accumulates semi-volatile compounds, hence the column needs to be baked out periodically to maintain optimal performance. Data collection and integration was conducted by PE Nelson 1020GC Plus software. During each analysis run, nine TFA standards, ranging from 1 to 500 ng, were run in duplicate. The standards consist of 1 mL of the TFA standard and 3 mL of 9 M sulfuric acid. New sets of standards were prepared from the stock solutions for each analysis run. Since the detector response was linear for TFA levels from 1 to 50 ng, a linear regression was used for the standard curve. Above 50 ng of TFA, the detector response became curvilinear, and a second-order polynomial was used for the standard curve. The detector response to TFA levels above 500 ng was deemed unreliable. When very high TFA levels are known to exist in a sample through dosing or previous analysis, the injection volume is decreased to 0.066 mL, instead of 0.33 mL, to prevent detector saturation and bring the response back within the detector range. Two additional high standards (1000 and 2000 ng) replaced the two lowest standards during these analysis runs. To confirm the presence of TFA in the plant samples, nine previously injected samples, three from each species grown in 830 µg/L of TFA, were analyzed by a Hewlett-Packard 5790A Gas Chromatograph coupled to a 5970A Mass Selective Detector (GCMSD). Since some of the headspace vapor was consumed by the previous analysis, a set of previously injected standards was used to quantitatively calibrate these repeat samples, thus the vapor loss was accounted for between these analyses. The analytical column was a 60-m DB-1 with a 1-µm stationary phase (J&W Scientific, Folsom CA). The temperature program started at 60 °C and ramped at 5 °C/min to 100 °C. This was then followed by a 25 °C ramp to 250 °C to bake out the column. The GC-MSD conditions were the following: helium carrier gas at 16.6 cm/s, source temp of 280 °C, and ionizing energy of 70 eV. The samples were thermostated in an oven at 50 °C for 1 h prior to the manual injection of 0.5 mL of the headspace vapor onto the column. Three ions, namely m/z 59, 69, and 99, were monitored in selective ion monitoring-electron impact (SIM-EI) mode to determine the presence and abundance of methyl TFA in the sample.25 Under these conditions, MTFA elutes at 7.4 min. Soil Analysis. The soil analysis procedure is a simplified version of the plant analysis with some minor modifications. Since soil does not actively concentrate TFA like plants, a larger sample (1 g) of soil is needed to obtain detectable TFA. The soil was added to the HS-GC vial along with 4 mL of the 75% sulfuric acid (9 M)/25% methanol derivatization solution. For soil samples that contained a significant quantity of organic matter, the sample size was reduced to 0.5 g, and 0.4 g of sodium sulfate was added to the vial to help displace the TFA ions from the organic matter. The vial was then placed on an orbital shaker for 3 h at 400 rpm. After shaking, the sample was ready for HS-GC analysis. Standards ranging from 0.125 to 25 ng of TFA were used for calibration. The soil spiking procedure was similar to the plant spiking procedure in that 1 mL of water containing the TFA spike was added to the sample. The water was slowly evaporated from the soil in a mechanical convection oven set at 50 °C. The HS-GC conditions were similar to those for the plant analysis. The highly acidic nature of the samples causes relatively 4468
Analytical Chemistry, Vol. 71, No. 20, October 15, 1999
rapid column wear, so the column needs to be replaced every 6 months. The GC column used for plants was replaced with a 25 m × 0.32 mm ID Porabond Q column (Chrompack, Raritan, NJ). This column is a bonded version of the Poraplot Q, so it is more stable and durable than the older particle-based PLOT columns. The Porabond column is a little less retentive, so the gas pressures and flows were lowered to compensate. The carrier flow was set to 0.9 mL/min of nitrogen using a column head pressure of approximately 41 kPa. The vials were pressurized to 138 kPa for 3 min prior to sample injection. The sample injection time was increased to 0.5 min in order to deliver more headspace vapor to the column and increase sensitivity. These conditions caused approximately 2.5 mL of the headspace vapor to be delivered to the column instead of the normal 0.33 mL, thus increasing the sensitivity. The relatively high vial pressure combined with the longer sampling time increased detector response by approximately 8-fold. The larger and longer sample injection dramatically broadened all peaks in the chromatogram, including the MTFA peak. The GC oven program was changed to compensate for the long injection by using a cool initial temperature to thermally focus the MTFA on the front of the column followed by a steep temperature ramp that elutes MTFA from the column. The initial temperature was 50 °C, which was held constant for 1 min during injection. The temperature was then ramped at 10 °C/min to 120 °C where it was held constant for 10 min. Methyl TFA eluted during this 120 °C plateau. The oven was then ramped at 40 °C/ min to 250 °C, where the column was baked out for 15 min. The early peaks in the chromatogram were still distorted due to the large injection, but the MTFA eluted as a relatively sharp peak at 13.4 min using this modified program (Figure 2B). These conditions were optimized for maximum sensitivity. Compared with the isothermal program, the ramp program gives less separation between the MTFA peak and other peaks, hence there is an increased potential for interferences. This program is vulnerable to gas contamination condensing in the cool column, so a pure gas supply is essential for it. Plant samples can also be run using these conditions, but the analysis runs are significantly longer, and the increase in sensitivity is typically unnecessary for the plant samples. Conversely, soil or water samples with very high TFA concentrations can use the time-optimized isothermal program. Water Analysis. The water analysis procedure was almost identical to the soil analysis technique. Fifteen milliliters of the sample water was placed in the sample vial along with 1 mL of 0.5 M ammonium carbonate, which raised the sample pH to approximately 9, thus ensuring that TFA remained as a salt during the evaporation step. The sample was evaporated to dryness in a mechanical convection oven at 90 oC for 8-14 h. The salts, including TFA, and organic material in the sample remained in the dry vial. Another 15 mL of sample, along with another 1 mL of ammonium carbonate, was added to the vial for a second evaporation cycle, thus increasing sample volume. Once the sample had evaporated to dryness, the derivatization solution (4 mL of 75% H2SO4/25% methanol) was added to the vial. The sample was then vortexed for 30 s in order to dissolve the salts on the vial walls, and it was then analyzed by HS-GC. The GC conditions were identical to the those for soil analysis procedure.
Table 1. Spike Recovery Trials Conducted Using Four Different Species of Plants TFA concentration (ng/g dry weight)
RSD (%)
253 757 2190
8.1 5.4 1.6
NA 100.8 96.9
106 585 1840
7.9 6.8 3.2
NA 95.8 86.7
101 603 2060
16.6 6.8 2.9
NA 100.4 98.0
144 751 2050
15.2 5.3 3.2 average recoveryb
NA 121.4 95.3 99.4%
herbaceous annual, monkeyflower (Mimulus gutlatus)a no spike 500 ng/g spike 2000 ng/g spike annual beard grass (Polypogon monspeliensis)a no spike 500 ng/g spike 2000 ng/g spike deciduous tree, black locust (Robinia pseudoacacia)a no spike 500 ng/g spike 2000 ng/g spike coniferous tree, Ponderosa pine (Pinus ponderosa)a no spike 500 ng/g spike 2000 ng/g spike a
spike recovery (%)
N ) 10 replicates per spike level. b Average of all 80 spike-recovery trials.
Saline samples, with chloride-ion concentrations greater than 75 mg/L, required a liquid-liquid cleanup extraction before they could be evaporated. The cleanup extraction was similar to a reported method in Wujcik et al.,26 except that the sample volume and solvents were scaled down so that the extraction could be conducted in a 50-mL test tube. Fifteen milliliters of the sample was added to a test tube along with 1.5 mL of 18 M sulfuric acid. The test tube was capped, shaken, and refrigerated at -10 °C for 15 min. In a second test tube, 0.25 mL of 0.5 M NaOH was added to 10 mL of Nanopure water. Once the first test tube had been chilled, 10 mL of ethyl ether was added to the test tube along with enough sodium sulfate to saturate the solution. The test tube was then shaken for 30 s. The layers were allowed to separate, and the ether was pipetted from the first test tube to the second. Two additional ether washes were conducted. Once all the ether was in the second test tube with the base, the test tube was shaken for approximately 1 min. Once the layers separated, the water solution was removed, along with 1-2 mL of ether, and was placed in the vial. The sample was heated to 50 °C for 3 h in order to drive off the ether that was dissolved within the water without boiling or bumping. If an appropriate solvent oven is unavailable, the ether can be removed from the water by placing the sample under vacuum for 3 h. After this time, the temperature was increased to 90 °C, and evaporation proceeded as normal. RESULTS AND DISCUSSION Plant Results. The spike-recovery trials showed a mean recovery of 99.4%, while individual trials’ recoveries ranged from 86.7 to 121.4% (Table 1). Since plants grown in solutions with no added TFA still showed detectable TFA concentrations, this background concentration was subtracted from the spike-recovery value. The average recovery for 24 reagent spikes was 102.4%, while individual recoveries ranged from a low of 87.2% to a high of 119.0%. The minimum detectable limit (MDL) and minimum quantifiable limit (MQL) were determined from the value and standard deviation (SD) of the lowest standard used for that sample type. The lowest standard in the plant analyses, which was always
Table 2. Plant Uptake of TFA from Hydroponic Solutions
monkeyflower (Mimulus gutlatus)a no TFA added 8.3 µg/L TFA 83 µg/L TFAb 830 µg/L TFAc rice (Oryza sativa)a no TFA added 8.3 µg/L TFA 83 µg/L TFAb 830 µg/L TFAc annual beard grass (Polypogon monspeliensis)a no TFA added 8.3 µg/L TFA 83 µg/L TFAb 830 µg/L TFAc
TFA concentration (ng/g dry weight)
RSD (%)
2.52 × 102 2.17 × 103 3.78 × 104 2.99 × 105
8.2 3.4 3.7 6.6
1.26 × 102 2.10 × 103 1.92 × 104 1.39 × 105
7.7 4.4 7.1 5.7
1.05 × 102 1.44 × 103 1.38 × 104 1.26 × 105
7.9 5.8 3.5 7.4
a N ) 10 laboratory replicates for each dosing level. b 0.01 g of plant material used for sample. c 0.01 g of plant material used and a sample injection volume of 0.066 mL.
detected, had an average reported value of 0.95 ng and a SD of 0.24 ng. Therefore, the MDL (average + 3SD) was 1.7 ng and the MQL (average + 6SD) was 2.4 ng. For a typical 0.05-g dry weight plant sample, these values correspond to an MDL of 34 ng/g and an MQL of 48 ng/g. In all the trials, the relative standard deviation (RSD) was less than 17%, and the RSD decreased with higher concentrations (Table 1). Plants hydroponically dosed with TFA for 36-48 days were analyzed to determine if the technique was able to extract TFA that the plant assimilated. The results showed increasing plant TFA concentration as the dosing water TFA concentration increased (Table 2). Furthermore, the plant TFA concentration increased approximately 10-fold along with the dosing level, and RSDs were less than 10%. These trials showed that the methodology was effective, but they did not provide much information about the TFA uptake function of the plants because all the plants in Analytical Chemistry, Vol. 71, No. 20, October 15, 1999
4469
Table 3. TFA Concentrations in Rice as Analyzed by Wet Weight and Dry Weight Proceduresa
Table 4. Spike Recovery Trials Conducted Using Four Different Soil Types
rice (Oryza sativa) dry-weight wet weight % difference grown in analysis RSD analysis RSD after wet/dry hydroponic solutions (ng/g)b (%) (ng/g)b (%) adjustment no TFA addedc 10 µg/L sodium TFA addedc 100 µg/L sodium TFA addedd
126 2100
7.7 4.4
19 200
7.1
40.5 636 4880
23.3 9.2
+15.2 +8.6
8.6
-8.9
a Wet plant material consists of 72.1% water and 27.9% dry material, therefore dry wt concentrations ) wet wt concentrations/0.279. These values were determined by freeze-drying whole plants. b N ) 10 replicates for each trial. c 0.05 g of dry plant material used for sample and 0.5 g of wet plant used, respectively. d 0.01 g of dry plant material used for sample and 0.1 g of wet plant used, respectively.
each group were homogenized into one powder to reduce variability. In this limited experiment, the concentration of TFA in the plant, in units of ng/g dry weight, was between 100 and 450 times higher, depending on species, than the water concentration (ng/g) of TFA. If the plant TFA concentration is expressed as ng/g wet weight, then plant concentrations are approximately 35 to 130× higher than that of the water solution. The comparison of the rice samples processed by wet-weight and dry-weight methods showed that the sample preparation did not affect the results (Table 3). The percent difference between the two methods, after adjusting for moisture content, was between +15.2 and -8.9%. The use of wet-weight sample preparation shortened analysis time and reduced possible contamination by bypassing the drying and grinding steps. Three trials using 10 aliquots of the 8.3 µg/L dosed beard grass tested the effect of a total digestion time of 24, 48, or 72 h. The digestion time was defined as the time during which the sample was exposed to the sulfuric acid plus the time after the addition of methanol. The mean values ((SD) for three trials were 1510 ( 21, 1440 ( 84, and 1490 ( 31 µg/g dry weight, respectively. To be conservative, a two-day digestion is suggested, particularly for tough samples such as pine needles. If the digestion is still inadequate after 2 days, then additional time can be taken to digest the plant without MTFA escaping from the vial. The 9 samples that were reanalyzed by the GC-MSD showed fairly consistent results with the first analysis. On average, the discrepancy between the two analyses was 18.4%, with the GCMSD giving slightly higher values than the HS-GC. This probably resulted from imperfect manual injections on the GC-MSD compared with the automated HS-GC. The peaks in the samples eluted at the same time as the TFA standard and had the characteristic fragmentation pattern for the derivatized TFA, thus confirming the presence and approximate abundance of TFA in the dosed plant samples. The GC-MSD response was linear over the four-point calibration curve (R2 > 0.99) using the previously injected standards. The previously injected standards showed consistently lower (average ) 49%) MTFA than new standards run only on the GCMSD. This indicates that about half the sample was used in the previous analysis. Most of this loss was probably post-analysis leaking from the pressurized and punctured vial. The mass spectrometer could detect 250 ng of TFA in the 22-mL headspace vial, so plant samples containing 5000 ng/g of TFA could be 4470 Analytical Chemistry, Vol. 71, No. 20, October 15, 1999
TFA concentration (ng/g) purified sand a no spike 6.25 ng/g spike 25 ng/g spike agricultural soil, Reno, NVa no spike 6.25 ng/g spike 25 ng/g spike playa soil, Stead, NVa no spike 6.25 ng/g spike 25 ng/g spike humic forest soil, Tahoe National Forest, CAa,c no spike 6.25 ng/g spike 25 ng/g spike
RSD (%)
spikerecovery (%)