Anal. Chem. 2005, 77, 6353-6358
Matrix Effect-Free Analytical Methods for Determination of Perfluorinated Carboxylic Acids in Environmental Matrixes Charles R. Powley,*,† Stephen W. George,† Timothy W. Ryan,† and Robert C. Buck‡
DuPont Haskell Laboratories, P.O. Box 50, Newark, Delaware 19714-0050, and DuPont Chemical Solutions Enterprise, Barley Mill Plaza 23, P.O. Box 80023, Wilmington, Delaware 19880-0023
Perfluorocarboxylic acids (PFCAs) are persistent chemicals that have been found widely in the environment. Their accurate determination in environmental matrixes, particularly soil, sediment, and sludge, at low levels presents significant analytical challenges. The commercialization of electrospray interfaces for liquid chromatographymass spectrometric analysis facilitated analysis of PFCAs at low levels, but issues with quantitative analysis due to matrix suppression or enhancement still persist. The methods described in this study utilize simple and rapid sample purification procedures to remove matrix components sufficiently so that errors due to coeluting matrix peaks are negligible and recoveries of PFCAs are consistently and reproducibly quantitative. Extracts from solid samples (soil and sediment) and liquid bacterial sludge are purified using dispersive solid-phase extraction. Recovery values generally are in the 70-120% range, with limits of quantitation of 1 ppb. The method utilizes an extraction solvent previously shown to release and recover aged residues of PFCAs. A confirmatory method using two precursor to product ions is also provided and demonstrated. Perfluorinated carboxylic acids (F(CF2)nCOOH, where n ) 7-11, 13 (PFCAs)) have been reported in a wide array of environmental matrixes including biota, water, air, sludge, sediment, and soil.1-5 PFCAs are generally regarded as highly persistent substances, which highlights the urgent need for validated analytical methods to understand their sources, fate, and transport in the environment.6 Issues and concerns regarding analytical methods for determination of PFCAs and other perfluorinated substances in the environment were recently reviewed, * To whom correspondence should be addressed. E-mail: Chuck.Powley@ usa.dupont.com. † DuPont Haskell Laboratories. ‡ DuPont Chemical Solutions Enterprise. (1) Taves, D. R. Nature 1968, 217, 1050-1051. (2) Giesy, J. P.; Kannan, K. Environ. Sci. Technol. 2002, 36, 146A-152A. (3) Martin, J. W.; Whittle, D. M.; Muir, D. C. G.; Mabury, S. A. Environ. Sci. Technol. 2004, 38, 5379-5385. (4) Higgins, C. P.; Field, J. A.; Criddle, C. S.; Luthy, R. G. Environ. Sci. Technol. 2005, 39, 3946-3950. (5) Yamashita, N.; Kannan, K.; Taniyasu, S.; Horii, Y.; Okazawa, T.; Petrick, G.; Gamo, T. Environ. Sci. Technol. 2004, 38, 5522-5528. 10.1021/ac0508090 CCC: $30.25 Published on Web 09/03/2005
© 2005 American Chemical Society
noting contamination, calibration, recovery, and separation as recurrent difficulties to be overcome.7 Determination of PFCAs in environmental and biological matrixes is quite challenging, due to their lack of volatility for gas chromatographic (GC) analysis and the lack of a suitable chromophore for liquid chromatographic (LC) analysis using ultraviolet detection.7 The methyl esters of PFCAs can be formed by chemical derivatization; these compounds may then be determined by GC with mass spectrometric or electron capture detection.8 The commercialization of the electrospray interface for liquid chromatography-mass spectrometry analysis has made PFCA analysis much more sensitive and specific than previous approaches. Liquid chromatography with negative ion electrospray ionization and tandem mass spectrometry (LC/MS/MS) has been used to determine selected PFCAs in surface water and groundwater. Analysis was accomplished by direct injection9 or preconcentration on octadecylsilyl (C18) solidphase extraction (SPE) cartridges, followed by LC/MS/MS analysis.5,10-12 Recently, an alternative SPE concentration and purification approach using a polymeric sorbent was reported for determination of perfluorooctyl and perfluorononyl carboxylic acids in seawater.13 There are few reports of methods for determining PFCAs in solid environmental matrixes. Soil, sediment, and sludge are complex environmental matrixes where substantial interaction with chemical substances including adsorption and sequestration can confound the ability to achieve reliable (6) U.S. EPA. Perfluorooctanoic Acid (PFOA), Fluorinated Telomers; Request for Comment, Solicitation of Interested Parties for Enforceable Consent Agreement Development and Notice of Public Meeting. U.S. EPA Administrative Record OPPT-2003-0012-0001. U.S. Environmental Protection Agency, Washington, DC, 1999. (7) Martin, J. W.; Kannan, K.; Berger, U.; deVoogt, P.; Field, J.; Giesy, J. P.; Harner, T.; Muir, D. C. G.; Scott, B. F.; Kaiser, M.; Ja¨rnberg, U.; Jones, K. C.; Mabury, S. A.; Schroeder, H.; Simicik, M.; Sottani, C.; vanBavel, B.; Ka¨rrman, A.; Lindstro ¨m, G.; vanLeeuwen, S. Environ. Sci. Technol. 2004, 38, 248A-255A. (8) Moody, C. A.; Field, J. A. Environ. Sci. Technol. 1999, 33, 2800-2806. (9) Moody, C. A.; Hebert, G. N.; Strauss, S. H.; Field, J. A. J. Environ. Monit. 2003, 5, 341-345. (10) Hansen, K. J.; Johnson, H. O.; Eldridge, J. S.; Butenhoff, J. L.; Dick, L. Environ. Sci. Technol. 2002, 36, 1681-1685. (11) Moody, C. A.; Kwan, W. C.; Martin, J. W.; Muir, D. C. G.; Mabury, S. A. Anal. Chem. 2001, 73, 2200-2206. (12) Moody, C. A.; Martin, J. W.; Kwan, W. C.; Muir, D. C. G.; Mabury, S. A. Environ. Sci. Technol. 2002, 36, 545-551. (13) So, M. K.; Taniyasu, S.; Yamashita, N.; Giesy, J. P.; Zheng, J.; Fang, Z.; Im, S. H.; Lam, P. K. S. Environ. Sci. Technol. 2004, 38, 4056-4063.
Analytical Chemistry, Vol. 77, No. 19, October 1, 2005 6353
quantitative analysis. Higgins4 reported PFCA levels in sediment and sludge with recoveries between 53 and 95%. The lowest fresh and aged fortification recoveries were reported for higher chain length PFCAs such as perfluorododecanoic and -tetradecanoic acids. Perfluorooctanoic acid (PFOA) was determined in vacuum cleaner house dust by extraction using methanol followed by LC/MS/MS analysis.14 Methods for determination of PFOA in biodegradation studies in sludge and soil15-17 and in extracts from consumer articles have also been recently reported.18 Robust analytical methodology for the determination of a homologous series of PFCAs in matrixes such as soil, sediment, and sludge is needed to better understand their presence, fate, and transport in the environment. Although LC/MS/MS with electrospray ionization demonstrates excellent sensitivity and specificity for PFCAs in a wide variety of matrixes without the need for chemical derivatization, there are still challenges that must be overcome in order to obtain quantitative data that are sufficiently reliable for the purposes of monitoring and government regulation. If unpurified soil/sediment/sludge extracts are analyzed, the quantitative results are often adversely affected by the phenomena commonly referred to as matrix suppression and matrix enhancement.19 For example, Dinglasan16 and Wang17 reported microbial degradation studies where calibration standards had to be made up in control matrix extract, to compensate for matrix effects. Matrix suppression occurs in the electrospray interface when coeluting matrix components compete with the analyte for charge, thereby reducing the number of gas-phase ions available for detection. Conversely, if a matrix component facilitates the ionization process (e.g., by reducing surface tension), an enhancement is obtained.20 Ideally, stable isotope analogues of the analytes can be used as internal standards to compensate for matrix effects, but they are of limited availability due to the cost of their synthesis.7 Purification using solid-phase extraction cartridges is also possible, but this can be time-consuming and effective procedures often do not apply to analytes with a wide range of physical properties. However, Anastassiades et al.21 developed a “dispersive solid-phase extraction” procedure whereby an aliquot of the crude extract is simply mixed with milligram quantities of bulk sorbent to remove matrix components and leave the analytes in solution. We modified this procedure to apply to PFCAs. Any method used to generate analytical data for the purposes of monitoring and risk assessment should release and recover residues from field-weathered samples, be documented and validated for all analytes and matrixes covered, and demonstrate acceptable accuracy and precision for fortifica(14) Moriwaki, H.; Takat, Y.; Arakawa, R. J. Environ. Monit. 2003, 5, 753-757. (15) Wang, N.; Szostek, B.; Folsom, P. W.; Sulecki, L. M.; Capka, V.; Buck, R. C.; Berti W. R.; Gannon, J. T. Environ. Sci. Technol. 2005, 39, 531-538. (16) Dinglasan, M. J. A.; Ye, Y.; Edwards, E. A.; Mabury, S. A. Environ. Sci. Technol. 2004, 38, 2857-2864. (17) Wang, N.; Szostek, B.; Buck, R. C.; Folsom, P. W.; Sulecki, L.; Powley, C.; Berti, W.; Gannon, J. T, Society of Environmental Toxicology & Chemistry, 2004 World Congress presentation, Portland, OR, November 2004, Abstr. PW044. (18) Mawn, M. P.; McKay, R. G.; Ryan, T. W.; Szostek, B.; Powley, C. R.; Buck, R. C. Analyst 2005, 130, 670-678. (19) Clarke, S. D.; Hill, H. M.; Noctor, T. A. G.; Thomas, D. Pharm. Sci. 1996, 2, 203-207. (20) Enke, C. G. Anal. Chem. 1997, 69, 4885-4893. (21) Anastassiades, M.; Lehotay, S. J.; Sˇ tajnbaher, D.; Schenk, F. J. J. AOAC Int. 2003, 86, 412-431.
6354
Analytical Chemistry, Vol. 77, No. 19, October 1, 2005
tions into the actual sample matrixes. In addition, it should be specific enough to avoid most false positive results and include a confirmatory procedure (such as a second MS/MS transition) to eliminate virtually all false positives.22 The method reported in this paper was developed to fulfill these basic criteria for a full homologous series of PFCAs, in addition to removing matrix effects using dispersive solid-phase extraction as a rapid and simple purification procedure and represents a robust, reliable method for the quantitative determination of PFCAs in solid environmental matrixes. EXPERIMENTAL SECTION Standards and Reagents. Perfluorohexanoic acid (C6), perfluoroheptanoic acid (C7), perfluorooctanoic acid (C8, PFOA), perfluoronanoic acid (C9), perfluorodecanoic acid (C10), perfluoroundecanoic acid (C11), perfluorododecanoic acid (C12), and perfluorotetradecanoic acid (C14) were all 97% purity or better and were purchased from Fluorochem (West Columbia, SC) or Sigma-Aldrich (Milwaukee, WI). The internal standard was dual 13C perfluorooctanoic acid that was custom-synthesized by DuPont. HPLC-grade acetonitrile and methanol were purchased from EM Science (Gibbstown, NJ). Bulk Envi-Carb sorbent (100 m2/g, 120/400 mesh) was purchased from Supelco (Bellefonte, PA). Safety precautions were standard for handling chemicals and flammable solvents. Appropriate personal protective equipment was worn at all times. Control Matrixes. Soil was obtained from Rochelle, IL, and Newark, DE. Sediment was obtained from Landenburg, PA. Bacterial sludge samples were grown within DuPont and were delivered in growth media solution. This was an activated sludge used in industrial waste treatment processes and is similar to that used in municipal systems. Sample Preparation. Five grams of solid sample (soil or sediment) was weighed into a 50-mL polypropylene disposable centrifuge tube. Fortifications of the analytes were then made by adding an appropriate volume of a 100 ng/mL solution in acetonitrile. Two milliliters of 200 mM NaOH in water was added; the samples then were allowed to soak for 30 min. Twenty milliliters of methanol and 250 µL of 100 ng/mL internal standard (dual 13C-PFOA) in acetonitrile were added to the sample tube. The sample tube was placed on a wrist-action shaker set at maximum deflection and mixed for 30 min. After the extraction was complete, 200 µL of 2 M HCl in water was added and the sample was shaken briefly by hand to neutralize the solution. The solids were allowed to settle for at least 30 min. Alternatively, the sample could be centrifuged at ∼3000 rpm for 20 min to clarify the supernatant sufficiently to remove an aliquot. The procedure for extraction of sludge samples was similar; 5 g of liquid sludge was weighed into a 50-mL polypropylene disposable centrifuge tube. Fortifications of the analytes and internal standard were then made by adding an appropriate volume of a 100 ng/mL solution in acetonitrile. Twenty milliliters of methanol and 250 µL of 100 ng/mL internal standard (dual 13C-PFOA) in acetonitrile were added to the sample tube. The sample tube was placed on a wrist-action shaker set at maximum (22) Li, L. Y. T.; Campbell, D. A.; Bennett, P. K.; Henion, J. Anal. Chem. 1996, 68, 3397-3404.
Figure 1. LC/MS/MS chromatogram of instrument background.
deflection and mixed for 30 min. The solids were allowed to settle for at least 30 min. Alternatively, the sample could be centrifuged at ∼3000 rpm for 20 min to clarify the supernatant sufficiently to remove an aliquot. Approximately 25 mg of Envi-Carb graphitized carbon adsorbent was added to a 1.7-mL disposable polypropylene microcentrifuge tube. Fifty microliters of glacial acetic acid was added directly to the sorbent. A 1-mL aliquot of the sample supernatant was added to the microcentrifuge tube, the tube was capped, and the contents were mixed using a vortex mixer. The sample was then centrifuged for 30 min at 10 000 rpm in a microcentrifuge. A 500-µL aliquot of the supernatant was pipetted into an HPLC autosampler vial, 500 µL of Nanopure water was added, and the contents mixed. For a matrix spike, 500 µL of a control sample prepared as above, 50 µL of 100 ng/mL internal standard solution, 100 µL of 100 ng/mL fortification solution, and 350 µL of water were combined into an autosampler vial. The sample was then ready for LC/MS/MS analysis. HPLC/MS/MS Analysis. Purified sample extracts were analyzed using a Quattro-Micro tandem mass spectrometer (Waters Corp.) coupled to an Agilent 1100 liquid chromatograph with an electrospray interface operating in the negative ion mode. The HPLC column used was Zorbax Rx-C8, 15 cm × 2.1 mm i.d., 5-µm packing (Agilent Technologies, Little Falls, DE). A Luna C182 column, 3 cm × 4.6 mm i.d., 3-µm packing (Phenomenex, Torrence, CA) was inserted in the HPLC between the pump and injector to remove any fluorochemicals originating from poly(tetrafluoroethylene) instrument components. The mobile phase consisted of acetonitrile and 0.15% (v/v) acetic acid in
water, programmed as follows:
time (min)
% acetonitrile
0 1 1.1 5 8 9 9.1 14
10 10 45 45 85 85 95 95
Injection volumes of 100-200 µL were used, with a flow rate of 0.4 mL/min and a column temperature of 40 °C. The mass spectrometer parameters were optimized to give a maximum response for C6. These parameters were also found to be optimum for the higher acids. An electrospray interface was used, operated in the negative ion mode. The precursor to fragment transitions monitored were C6, m/z 313 > 269, 313 > 119; C7, m/z 363 > 319, 363 > 169; C8, m/z 413 > 369, 413 > 169; 13C C8 (internal standard), m/z 415 > 370; C9, m/z 463 > 419, 463 > 169; C10, m/z 513 > 469, 513 > 169; C11, m/z 563 > 519, 563 > 169; C12, m/z 613 > 569, 613 > 169; and C14, m/z 713 > 669, 713 > 169. In each case, the primary transition (bold face type) used for screening and quantitation is the first one listed, and corresponds to (M - CO2)-. The secondary transition was used for confirmatory purposes. The collision voltage was 9 eV for the primary transition and 12 eV for the secondary transition. The collision gas was argon at a pressure of 0.0036 mbar. The cone and capillary voltages were Analytical Chemistry, Vol. 77, No. 19, October 1, 2005
6355
17 and 3500 V, respectively. The source and desolvation temperatures were 120 and 350 °C. Calibration and Validation. Calibration standards were made up in 1:1 methanol/water. The concentration range of the five to six calibration standards prepared was 0.1-10 ng/mL. Each calibration standard also contained the internal standard at a concentration of 1 ng/mL. The calibration range was selected to ensure that the least concentrated standard in each set corresponded to ∼50% of the 1 ppb limit of quantitation (LOQ), and the highest standard corresponded to 120% or more of 10×LOQ, or 10 ppb. Calibration curves were constructed by using the QuanLynx software of the LC/MS/MS instrument to perform linear regression (1/x weighting) of plots of (peak area/internal standard area) versus (standard concentration/internal standard concentration). Each validation set consisted of a reagent blank (prepared by taking an appropriate volume of the extracting solution through the entire procedure), an unfortified control sample, five samples fortified at 1 ppb (the LOQ), and five samples fortified at 10 ppb (10×LOQ). In addition, aliquots of the control sample crude and purified extracts were fortified at a concentration corresponding to 10 ppb immediately prior to analysis in order to demonstrate the absence of matrix effects. RESULTS AND DISCUSSION Instrumental Background Noise. Many HPLC systems contain components and lubricants made from poly(tetrafluoroethylene) (PTFE) in their solvent delivery systems. PTFE can contain traces of PFCAs and other perfluorinated compounds that can leach into the mobile phase. Previous reports23 have mentioned background peaks observed in the C8 and C9 channels, eluting very closely to the analytes. Figure 1 is a chromatogram of the instrument background obtained from our system when no injection was made. There are peaks eluting on the C6-C10 channels that are significant enough to interfere at the lowest concentration ranges (0.1-0.2 ng/mL) that we studied. No response was observed in the internal standard channel, which indicates that this background is not due to carryover from injections of samples, all of which contain the internal standard. The insertion of a mini-HPLC column (3 cm × 4.6 mm i.d., 3-µm packing) between the pump and injector appears to have solved this problem, as interfering background peaks were no longer observed on any of the channels. Instead of focusing on the analytical column, the impurities leaching from the instrument components focus at the head of the guard column. Due to the more retentive nature of the sorbent (C18 vs C8) and the similar column volume, the instrument peak elutes well after all of the analytes of interest, when data acquisition has been stopped. Calibration. Calibration curves were generally linear with correlation coefficients (r2) of 0.999 or better. The value of the intercept was generally 10% or less of the lowest standard area ratio. Therefore, linear regression (1/x weighting) was justified as the method of choice for generating an equation to calculate concentrations from peak areas. Limit of Quantitation. The LOQ for each matrix was defined as the lowest fortification level validated, which was 1 ppb for soil, (23) Martin, J. W.; Smithwick, M. M.; Braune, B. M.; Hoekstra, P. F.; Muir, D. C. G.; Mabury, S. A. Environ. Sci. Technol. 2004, 38, 373-380.
6356
Analytical Chemistry, Vol. 77, No. 19, October 1, 2005
sediment, and sludge. The LOQ corresponds to a ∼10:1 peak-topeak/signal-to-noise ratio for the least responsive analyte (C14). Controls. The matrixes collected had levels (469/ 513>169)
C11 (563>519/ 563>169)
C12 (613>569/ 613>169)
C14 (713>669/ 713>169)
Table 3. Results of Confirmatory Analysis species determined (m/z) C6 (313>269/ 313>116)
C7 (363>319/ 363>169)
C8 (413>369/ 413>169)
average intensity ratio (n ) 10) (30% window of acceptability
1.53 1.07-1.99
Standards (0.25-5 ng/mL) 0.68 0.73 1.34 0.48-0.88 0.51-0.95 0.94-1.74
1.57 1.10-2.04
2.91 2.04-3.78
0.74 0.52-0.96
3.77 2.64-4.90
intensity ratio at 1 ppb intensity ratios at 10 ppb
1.60, 1.33 1.52
0.65, 0.63 0.73
Illinois Soil 0.64, 0.66 1.45, 1.10 0.67 1.35
2.00, 1.94 1.77
3.44, 2.79 3.04
0.65, 0.96 0.71
3.53 3.76
intensity ratio at 1 ppb intensity ratios at 10 ppb
1.57, 1.22 1.68, 1.47
Pennsylvania Pond Sediment 0.61, 0.61 0.61, 0.74 1.32, 1.09 0.65, 0.70 0.65, 0.79 1.23, 1.19
1.55, 1.59 1.65, 1.97
2.91, 3.05 2.89, 3.42
0.94, 0.71 0.71, 0.70
3.05 3.94, 3.64
intensities obtained for the standards, the detection is confirmed.22 The results of confirmatory analysis performed on some of the procedural recovery study samples are shown in Table 3. These samples were 1 and 10 ppb fortified control soil from Illinois and pond sediment that had been taken through the entire extraction and purification procedure. The ratios for the fortified samples at 1 and 10 ppb fall within the (30% window for each analyte. However, some of the 1 ppb (LOQ) samples were not confirmed
6358 Analytical Chemistry, Vol. 77, No. 19, October 1, 2005
C9 (463>419/ 463>169)
due to the absence of a second precursor to product transition. If confirmation is necessary at LOQ or at levels between the LOD and LOQ, it may be necessary to concentrate the samples or inject larger volumes to achieve quantifiable signals for both transitions. Received for review May 10, 2005. Accepted August 4, 2005. AC0508090
