Anal. Chem. 2004, 76, 5518-5522
Determination of Perchlorate Anion in Foods by Ion Chromatography-Tandem Mass Spectrometry Alexander J. Krynitsky,* Richard A. Niemann, and David A. Nortrup
Center for Food Safety and Applied Nutrition, U.S. Food and Drug Administration, 5100 Paint Branch Parkway, College Park, Maryland 20740-3835
A rapid, sensitive, and specific method was developed for determining perchlorate anion in lettuce, cantaloupe, bottled water, and milk. A test portion of chopped crop homogenate was extracted with diluted nitric acid and filtered. Milk proteins were precipitated with acetonitrile, and the supernatant, after centrifugation, was cleaned up on a graphitized carbon solid-phase extraction column. Water samples were analyzed directly. All test solutions were syringe filtered and mixed with an 18O4-labeled perchlorate internal standard before ion chromatography-tandem mass spectrometry. A strong anion exchange column eluted with 100 mM ammonium acetate in 50:50 (v/v) acetonitrile/water was interfaced via electrospray ionization to a triple stage quadrupole mass spectrometer operated in the negative ion mode. The labeled internal standard corrected for any sample matrix effects on measured signals. Four parent-to-product ion transitions, for loss of oxygen, were monitored for native and 18O -labeled perchlorate anion, respectively: 35Cl-per4 chlorate, m/z 99 f 83 and 107 f 89; 37Cl-perchlorate, m/z 101 f 85 and 109 f 91. The limit of quantitation was 1.0 µg/kg in lettuce, 2.0 µg/kg in cantaloupe, 0.50 µg/L in bottled water, and 3.0 µg/L in milk. Native perchlorate was recovered from fortified test portions in the range 93-107% for lettuce, 107-114% for cantaloupe, 100-115% for bottled water, and 99-101% for milk. Perchlorate anion is an environmental contaminant usually associated with the storage, manufacture, and testing of solid rocket motors, which use ammonium perchlorate as an oxidizer.1-3 One source of perchlorate contamination is the removal and recovery of propellant rocket fuel from solid rocket motors, which can result in wastewater that contains ammonium perchlorate. Perchlorate infiltrates the watersheds through a variety of mechanisms, such as leaching and groundwater recharge. This infiltration can occur in water supplies of several regions, such as the southwestern United States. Potential health effects associated with the perchlorate anion include disruption of the thyroid * Corresponding author. Phone: 301-436-2098. Fax: 301-436-2632. E-mail:
[email protected]. (1) Urbansky, E. T. Biorem. J. 1998, 2, 81-95. (2) Urbansky, E. T.; Achock, M. R. J. Environ. Manage. 1999, 56, 79-95. (3) Magnuson, M. L.; Urbansky, E. T.; Kelty, C. A. Anal. Chem. 2000, 72, 2529.
5518 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
function by competitively inhibiting iodide transport.4 The resulting malfunction of the Na+-I- symporter5 can reduce thyroid hormone production6 and can impair development. Pregnant women, children, and people with compromised thyroid functions are particularly at risk.7 Recently, the state of California published a final Public Health Goal of 6 µg/L for perchlorate in drinking water.8 About 90% of the nation’s winter lettuce is grown in the Imperial Valley of California and Yuma, AZ, which are irrigated with Colorado River water contaminated with perchlorate.9 Thus, it is conceivable that public exposure to perchlorate could spread nationwide from these two growing areas. Particularly relevant for young children is exposure to perchlorate in milk. Dasgupta7 reported 1.7-6.4 µg/L perchlorate in all seven whole milk samples they collected from local supermarkets in Lubbock, TX. To assess exposure to perchlorate from the food supply, the U.S. Food and Drug Administration (FDA) developed a sensitive, specific, and rapid instrumental method to provide analytical data. Food analyses have a higher likelihood of false positive results when compared to simpler matrixes such as water, especially using methods not entailing mass spectrometry. Methods employing conductivity detection (CD) lacked perchlorate specificity and so required more laborious cleanup.7,10 Methods based on highly sensitive and specific mass spectrometry3,11-13 have been successful for determination of perchlorate anion in water and soil without relying on an isotopically labeled internal standard. However, in electrospray ionization mass spectrometry (ESI-MS) of extracts prepared from more chemically complex samples such as food, matrix effects often suppressed or enhanced the quantitation signal, possibly causing large errors in quantitation if signal was (4) Clewell, R. A.; Merrill, E. A.; Yu, K. O.; Mahle, D. A.; Sterner, T. R.; Mattie, D. R.; Robinson, P. J.; Fishert, J. W.; Gearhart, J. M. Toxicol. Sci. 2003, 73, 235-255. (5) Dohan, O.; De la Vieja, A.; Paroder, V.; Riedel, C.; Artani, M.; Reed, M.; Ginter, C. S.; Carrasco, N. Endocrinol. Rev. 2003, 24, 48-77. (6) Greer, M. A.; Goodman, G.; Pleus, R. C.; Greer, S. E. Environ. Health Perspect. 2002, 110, 927-937. (7) Dasgupta, P. K.; Kirk, A. B.; Smith, E. E.; Tian, K.; Anderson, T. A. Environ. Sci. Technol. 2003, 37, 4979-4981. (8) Public Health Goal for Perchlorate in Drinking Water; Office of Environmental Health Hazard Assessment; California Environmental Protection Agency: Sacramento, CA, March 2004. (9) Bustilla, M. Colorado River Taint [Perchlorate] Worries Some Officials. Los Angeles Times, February 2, 2003, p A1. (10) Ellington, J. J.; Evans, J. J. J. Chromatogr., A 2000, 898, 193-199. (11) Winkler, P.; Minter, M.; Wiley, J. Anal. Chem., 2004, 76, 469-473. (12) Urbansky, E. T. Crit. Rev. Anal. Chem. 2000, 30, 311-343. (13) Koester, C. J.; Beller, H. R.; Halden, R. U. Environ. Sci. Technol. 2000, 34, 1862-1864. 10.1021/ac049281+ Not subject to U.S. Copyright. Publ. 2004 Am. Chem. Soc.
Published on Web 08/05/2004
calibrated against pure standard solutions. Overcoming matrix effects through additional extract cleanup and calibration by standard additions would considerably lengthen ESI-MS methodology. Alternatively, the use of a labeled perchlorate anion as an internal standard and tandem mass spectrometry (MS/MS) should compensate for matrix effects without compromising analytical speed. We describe in this paper an ion chromatography (IC)-MS/ MS method that incorporates an 18O4-labeled perchlorate internal standard to correct for matrix effects on measured signals. To the best of our knowledge, this was the first time that an isotopically labeled internal standard was used for determination of perchlorate anion in foods. The labeled internal standard greatly simplified extract preparation and cleanup before IC-MS/MS analysis. High-moisture crops, such as lettuce and cantaloupe, were analyzed after a simple blending extraction and filtration. Extraction was not needed for any liquid samples. Milk protein was precipitated with acetonitrile, and the centrifuged supernatant was cleaned up on a graphitized carbon solid-phase extraction (SPE) column. Bottled water was analyzed directly after syringe filtration. EXPERIMENTAL SECTION Reagents. A certified perchlorate analytical standard in water (997 ( 6 µg/mL) was obtained from Alltech (Deerfield, IL) as the potassium salt. The 18O4-labeled sodium perchlorate internal standard (IS) was 98+% pure from Icon Services Inc. (Mt. Marion, NY). Acetonitrile, HPLC grade, and nitric acid, 69-70%, were “Baker Analyzed” (J. T. Baker, Phillipsburg, NJ). HPLC-grade ammonium acetate (98%) was obtained from Fisher Scientific (Fair Lawn, NJ). High-purity water was prepared by passing laboratorydeionized water (reversed osmosis plus ion exchange) through a Millipore Milli-Q Ultrapure water purification system (Billerica, MA). IC-MS/MS Working Standards. Aqueous calibration standards were prepared at a constant IS concentration of 10 µg/L and the following concentrations of native perchlorate anion: 0.25, 0.50, 1.0, 2.0, 5.0, 10.0, and 100.0 µg/L. Apparatus. A Micromass Quattro Micro (Manchester, U.K.) triple stage quadrupole mass spectrometer was interfaced to an Agilent model 1100 HPLC (Palo Alto, CA) equipped with an autosampler (50-µL injection loop), binary pump, and column heater (35 °C). Waters (Milford, MA) IC-Pak Anion HR analytical (4.6 mm i.d × 75 mm) and Anion Guard-Pak (5 mm i.d. × 10 mm) columns were eluted with 100 mM ammonium acetate in 50:50 (v/v) acetonitrile/water at 300 µL/min. Milk samples were cleaned up on Supelclean ENVI-Carb (Supelco, Bellefonte, PA) SPE columns (500-mg bed of graphitized carbon in 6-mL tube) used with a Visiprep 24 (Supelco) SPE manifold. Sample Preparation. Lettuce and Cantaloupe. A 100-g test portion of chopped edible commodity was added to a Waring blender (or equivalent device) and homogenized for 6 min with 150 mL of 10 mM nitric acid. The contents plus two 20-mL dilute acid rinses of the blender jar were filtered under vacuum through a glass fiber filter (11 cm, GF no. 30, Schleicher&Schuell MicroScience, Inc., Riviera Beach, FL), and the final volume of extract was noted. An aliquot drawn into a syringe was filtered (0.45-µm Nylon) into a graduated concentrator tube (1.0 × 0.1 mL, Kontes, Vineland, NJ), which already contained 100 µL (via
syringe) of 0.10 µg/mL IS solution, until the 1.0-mL mark was reached. The contents of the tube were brought to the neck of the stoppered tube and were mixed by gentle rotation. Milk. Into a 50-mL polypropylene conical centrifuge tube was pipetted a 5.0-mL test portion of whole milk and 100 µL of 4.0 µg/mL IS solution. After adding 5 mL of water and 20 mL of acetonitrile, the capped tube was shaken briefly and then centrifuged at 2000 rpm for 5 min. Supernatant was passed through a preconditioned (6 mL of acetonitrile followed by 6 mL of water) ENVI-Carb SPE column, and eluant was collected in a 50-mL graduated glass centrifuge tube. The column was washed with 6 mL of water, which was combined with the eluant. Water was added to bring the final volume to 40 mL, and after mixing, ∼1.0 mL of solution was syringe-filtered for IC-MS/MS analysis. Bottled Water. A 100-µL aliquot of 0.10 µg/mL IS solution was pipetted into a concentrator tube, syringe-filtered bottled water was added up to the 1.0-mL mark, and the contents were carefully mixed. IC-MS/MS Analysis. Conditions for negative ion ESI MS/ MS were as follows: 3.25 kV capillary, 50 V cone, 1.0 V extractor, 0.0 V rf lens; 120 °C source, 350 °C desolvation; nitrogen gas flow rates of 60 L/h cone and 510 L/h desolvation; 2 × 10-3 mbar argon collision gas pressure, and 30 eV collision energy. Four multiple reaction monitoring (MRM) transitions were observed: 35Cl m/z 99 f 83, primary transition for native perchlorate (quantitation); 37Cl m/z 101 f 85, secondary transition for native perchlorate (confirmatory); 35Cl m/z 107 f 89, primary transition for labeled perchlorate (quantitation); and 37Cl m/z 109 f 91, secondary transition for labeled perchlorate (confirmatory). Instrument calibration was performed by analyzing standard solutions in which the labeled IS was always 10 µg/L and native perchlorate anion ranged from 0.25 to 100 µg/L. Relative response was calculated by dividing the peak area of the m/z 99 f 83 transition by the peak area of the m/z 107f 89 transition and was plotted against native perchlorate anion concentration. Linear regression coefficients and the relative response observed with a test solution were used for quantitation. To demonstrate that the instrument remained properly calibrated, a calibration standard was analyzed after every 20 samples. Additional quality assurance was provided by analysis of matrix fortifications, method blanks, and replicate samples. RESULTS AND DISCUSSION Lettuce and cantaloupe were extracted with dilute nitric acid initially for an IC-CD procedure14 to suppress ionization of ionizable coextractives that interfered with measurement of perchlorate conductivity. The addition of 10 mM nitrate, however, did not affect IC-MS/MS measurements as nitrate eluted well before perchlorate. Although isotopically labeled internal standards usually are added at the beginning of the extraction, in this method, the IS had to be added just before IC-MS/MS because the extract was split for parallel analysis by IC-CD. Perchlorate recoveries from fortified lettuce and cantaloupe shown in Table 1 suggest negligible loss of perchlorate to the extraction and filtration steps, thereby obviating the need to add the IS at the beginning. The IC-Pak Anion HR column proved to be rugged. Over a 6-month period in which 500 injections of standards and sample (14) Niemann, R. A., to be submitted for publication.
Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
5519
Table 1. Recoveries of Perchlorate Anion Fortified in Lettuce, Cantaloupe, Bottled Water, and Whole Milk
commodity lettuce
cantaloupeb bottled water
whole milkc whole milkd
fortification levela
average % recovery (n ) 3)
% RSD
1.0 2.0 10.0 20.0 100 10.0 20.0 100 0.5 1.0 5.0 10.0 5.0 10.0 20.0 5.0 10.0 20.0
107 92.6 101 94.5 103 107 112 114 115 104 109 99.7 98.9 99.7 101 94.7 97.6 101
13.3 3.4 8.1 2.6 6.0 0.66 11.4 2.2 9.9 8.2 4.6 6.8 5.0 8.2 6.3 22.6 8.2 1.8
a µg/kg for lettuce and cantaloupe, and µg/L for bottled water and whole milk. b Edible portion of cantaloupe. c Labeled internal standard added before cleanup. d Labeled internal standard added after cleanup.
extracts were made, average retention times differed by (2.0%. The choice of mobile phase was optimized for chromatographic peak shape and ESI efficiency. Severe peak tailing was observed if concentration of ammonium acetate in the mobile phase was 50 mM (instead of 100 mM). Keeping the proportion of acetonitrile high (50%) enhanced detection sensitivity by facilitating desolvation of the mobile phase. Figures 1 and 2 show MRM chromatograms of extracts of iceberg lettuce fortified at 2.0 µg/kg and a control. The chromatograms show adequate signals for both primary and secondary transitions of native perchlorate at 0.033 ng (equivalent to 2.0 µg/ kg) and no significant interferences for the control lettuce. The responses observed for the isotopically labeled perchlorate, which was added to the test solution, were from 0.50 ng injected. A software smoothing function was applied to the acquired data. Table 2 summarizes perchlorate levels found in samples from a limited survey. Most of the lettuce was the 2004 winter crop purchased in local supermarkets. Figure 3 shows the distribution of perchlorate concentrations across 19 lettuce samples. Perchlorate in 84% of these samples was e20 µg/kg. The cantaloupe samples were collected mostly from packing sheds near Yuma, AZ, in June 2003. A large proportion of perchlorate anion in cantaloupe was distributed in inedible rind and seeds (Table 2), and the edible portion incurred only about half the total amount available in the whole commodity. Determination and confirmation of low microgram per kilogram levels of perchlorate demonstrated sufficient method sensitivity and specificity to support risk assessment of exposure to perchlorate from crops. Bottled water, a chemically simple matrix, was analyzed directly after syringe filtration and IS addition. As with the other commodities, the recoveries of perchlorate anion in bottled water were acceptable (Table 1). Graphitized carbon,15 which removed milk pigment from supernatant after milk protein precipitation and centrifugation, rendered a water-clear eluant. Performing a milk recovery study, 5520 Analytical Chemistry, Vol. 76, No. 18, September 15, 2004
Figure 1. IC-MS/MS of iceberg lettuce fortified with 2.0 µg/kg perchlorate: (A) m/z 109 f 91 IS secondary transition; (B) m/z 107 f 89 IS primary transition; (C) m/z 101 f 85 native perchlorate secondary transition; and (D) m/z 99 f 83 native perchlorate primary transition.
however, was complicated by lack of a blank milk sample, and so recovery results in Table 1 were corrected for the average perchlorate concentration found in the controls. In so doing, perchlorate recovery from fortified milk was said to be complete. Recovery values were independent of whether the IS was added to the test portion before treatment or to the prepared test solution just prior to IC-MS/MS. Also, reagent blanks were analyzed together with milk and other matrixes in order to verify that carryover was not present. We obtained 25 whole milk samples from various parts of the country, and our findings in Table 2 were similar to those reported by others.7 For one raw (unprocessed) milk sample, we verified the concentration by three point standard additions (5.0, 10.0, and 20.0 µg/L) to the test solution prepared from a control portion. The slope of the standard additions curve was within 0.2% of the slope of the standard calibration curve, and the concentration of incurred perchlorate by standard additions was within 4% of that found without standard additions. This showed that the isotopically labeled internal standard mitigated matrix effects. The distribution of perchlorate concentrations within 2 µg/kg intervals above the limit of quantitation (LOQ), displayed as a histogram in Figure 4, was virtually constant in the range of 3-9 µg/kg, which included 80% of the samples. No sample had perchlorate above 12 µg/kg. We compared our milk method to an independent MS/MS method by analyzing split samples supplied by an independent (15) Anastassiades, M.; Lehotay, S. J.; Stajnbaher, D.; Schenck, F. J. J. Assoc. Off. Anal. Chem. Int. 2003, 86, 412-431.
Figure 3. Distribution of perchlorate concentration found in 19 lettuce samples.
Figure 2. IC-MS/MS of iceberg lettuce control: (A) m/z 109 f 91 IS secondary transition; (B) m/z 107 f 89 IS primary transition; (C) m/z 101 f 85 native perchlorate secondary transition; and (D) m/z 99 f 83 native perchlorate primary transition. Table 2. Concentration of Perchlorate Anion Found in Samples commodity
n
concn rangea
median
source (state)
lettuceb cantaloupec cantalouped milk
19 11 10 25
