Isotope Dilution Analysis of Bromate in Drinking Water Matrixes by Ion

ion chromatography-inductively coupled plasma mass spectrom- etry (IC-ICPMS) for the detection of bromate in drinking water matrixes. Creed et al.24,2...
0 downloads 0 Views 70KB Size
Anal. Chem. 1999, 71, 722-726

Isotope Dilution Analysis of Bromate in Drinking Water Matrixes by Ion Chromatography with Inductively Coupled Plasma Mass Spectrometric Detection John T. Creed* and Carol A. Brockhoff

National Exposure Research Laboratory, Microbiological and Chemical Exposure Assessment Research Division, U.S. Environmental Protection Agency, 26 West Martin Luther King Drive, Cincinnati, Ohio 45268

Bromate is a disinfection byproduct in drinking water which is formed during the ozonation of source water containing bromide. This paper describes the analysis of bromate via ion chromatography-inductively coupled plasma mass spectrometry. The separation of bromate from interferences such as bromide and brominated haloacetic acids is achieved using a PA-100 column in combination with a 5 mM HNO3 + 25 mM NH4NO3 mobile phase. Polyatomic ions are observed on masses 79 and 81 in a synthetic phosphate matrix and in ozonated drinking waters. These polyatomic ions have been tentatively identified as PO3+ and H2PO3+. These polyatomic ions do not interfere with the detection of bromate because phosphate elutes prior to bromate. A polyatomic ion is observed on mass 81 in a synthetic sulfate matrix and in ozonated drinking waters. This polyatomic ion has been tentatively identified as HSO3+ and does not interfere with the detection of bromate because sulfate elutes after bromate. Isotope dilution analysis produces a relative standard deviation (RSD) of ∼5% for both enriched isotopic additions at sample concentrations of 10 ng/g. The RSD associated with the direct analysis of bromate is 3.2% at sample concentrations of 10 ng/g. The bromate concentrations determined in ozonated drinking waters via isotope dilution analysis are within 10% of the concentrations determined via direct analysis for sample concentrations above 2 ng/g. The detection limit for the direct analysis of bromate via ICICPMS is 0.3 ng/g. The disinfection of drinking water with ozone is becoming more prevalent as the health risks associated with disinfection byproducts (DBPs) formed during disinfection via chlorination become better documented.1 Ozonation, as a disinfection process, has its own set of DBPs including brominated trihalomethanes, bromoacetic acids, bromoacetones, and bromoacetonitriles.2 These DBPs and bromate are formed from the ozonation of source water containing bromide.3,4 The bromide in the source water is oxidized * Corresponding author: (telephone) (513) 569-7833; (fax) (513) 569-7757. (1) Mughal, F. H. J. Environ. Pathol., Toxicol. Oncol. 1992, 11 (5, 6), 287292. (2) Siddiqui, M. S.; Amy, G. L. J. Am. Water Works Assoc. 1993, 85, 63.

722 Analytical Chemistry, Vol. 71, No. 3, February 1, 1999

to bromate (via an intermediate OBr-) by the ozone. Bromate is believed to be a cancer-causing agent with an estimated lifetime cancer risk of 1 in 104 at 5 ppb based on an average adult’s drinking water intake.5 The World Health Organization has established a guideline of 25 µg/L, and a lifetime cancer risk of 1 in 105 for 3 µg/L, for bromate.6 This carcinogenic risk would require the maximum contaminant level goal (MCLG) associated with the Safe Drinking Water Regulations for bromate be set at zero. USEPA method 300.0 is an ion chromatography/conductivity detection-based method for the analysis of bromate in drinking water with a method detection limit (MDL) of 20 µg/L.7 A revised method, USEPA method 300.1, utilizes a high-capacity column and a carbonate eluent system and has a detection limit of 1.3 ppb.8 The analysis of bromate by ion chromatography coupled with conductivity detection suffers from a near coelution of chloride with bromate.9,10 Recent advances in minimizing this interference have included a column-switching system11 and the use of a silver-containing cartridge as a pretreatment to remove the chloride.12,13 Gordon et al.14 used the oxidation of chlorpromazine by bromate as the basis for detection,15 but this spectrophotometric technique still required chloride removal prior to analysis. Bohme et al.16 coated a reversed-phase material with an (3) Haag, W. R.; Holgne, J. Environ. Sci. Technol. 1983, 17, 261. (4) VonGunten, U.; Hoigne, J. Environ. Sci. Technol. 1994, 28, 1234. (5) Fed. Regist. 1994, Part 136, 59, No. 145, 38709. (6) Guidelines for drinking water quality; WHO: Geneva, 1993; p 96. (7) EPA method 300.0. Methods for the Determination of Inorganic Substances in Environmental Samples, EPA/600/R93/100, 1993. (8) EPA method 300.1, Determination of Inorganic Anions in Drinking Water by Ion Chromatography EPA 600/R-98/118, 1997. (9) Hautman, D. P.; Bolyard, M. J. Chromatogr. 1992, 602, 65. (10) Joyce, R. J.; Dhhillon, H. S. J. Chromatogr., A. 1994, 671, 165. (11) Hautman, D. Analysis of Trace Bromate in Drinking Water Using Selective Anion Concentration and Ion Chromatography. Paper presented at the American Water Works Association Water Quality Technology Conference, Toronto, Canada, 1992, American Water Works Association, Denver, CO, 1993; p 993. (12) Joyce, R. J.; Dhillon, H. S. Part-Per-Billion Level Determination of Bromate in Ozonated Drinking Water Using Ion Chromatography. Paper presented at the International Ion Chromatography Symposium, Baltimore, MD, 1993. (13) Weinberg, H. J. Chromatogr., A. 1994, 671, 141. (14) Gordon, G.; Bubnis, B.; Sweetin, D.; Juo, C. Ozone Sci., Eng. 1994, 16, 79. (15) Gordon, G.; Bubnis, B. Ozone Sci., Eng. 1995, 17, 551. (16) Bohme, U.; Schmidt, W.; Dietrich, P. G.; Matschi, A.; Sacher, F.; Brauch H. J. Fresenius J. Anal. Chem. 1997, 357, 629. 10.1021/ac980663n Not subject to U.S. Copyright. Publ. 1999 Am. Chem. Soc.

Published on Web 12/18/1998

ionogenic agent for the measurement of bromate in drinking waters via UV detection. Charles et al.17 used electrospray and ion chromatography in combination with tandem mass spectrometry to achieve sub-ppb detection limits for bromate in water. Inoue et al.18 utilized postcolunmn derivatization for the detection of bromate via the formation of Br3. Heitkemper et al.19-21 used ICPMS as a detector for an ion chromatograph for the analysis of bromate in bread. Diemer et al.22,23 utilized isotope dilution and ion chromatography-inductively coupled plasma mass spectrometry (IC-ICPMS) for the detection of bromate in drinking water matrixes. Creed et al.24,25 have utilized an ICPMS detector for direct and isotope dilution analyses of bromate in the presence of brominated haloacetic acids using a nitrate-based eluent system. This system has a relatively low tolerance to high matrix anion concentrations because of the 5 mM HNO3 mobile phase and lowcapacity guard column used in the separation.25 The focus of this paper is 2-fold: to improve the matrix tolerance of the existing systems in order to improve its applicability to drinking waters with high ionic strength and to investigate the use of isotope dilution analysis for the determination of bromate in ozonated drinking waters by IC-ICPMS. The use of isotope dilution analysis provides a rapid, cost-effective, and accurate means of determining bromate below 1 ng/g in drinking waters samples. This procedure can incorporate preconcentration (via concentrator columns) which can compensate for losses due to column capacity limitations in high ionic strength waters. In addition, this investigation will include a comparison of direct and isotope dilution analysis via IC-ICPMS. EXPERIMENTAL SECTION Reagents. The 5 mM HNO3 was made from 624 µL of 1:1 HNO3 (Ultrex II, J. T. Baker Inc., Phillipsburg, NJ) diluted to 1 L. The NH4NO3 used in the mobile phase was ACS reagent grade (Fisher, Fairlawn, NJ). All dilutions were made using 18 MΩ water (Millipore, Bedford, MA). The bromate standards were made from sodium bromate (Fisher). The NaBrO3 standard did not contain detectable quantities of bromide and was verified against a secondary bromate standard (High Purity, Charleston, SC). The bromide standard was made from NaBr (Fisher). The methanol (17) Charles, L.; Pepin, D.; Casetta, B. Anal. Chem. 1996, 68, 2554. (18) Inoue, Y.; Sakai, T.; Kumagai, H.; Hanaoka, Y. Anal. Chim. Acta 1997, 346, 299. (19) Heitkemper, D. T.; Kaine, L. A.; Jackson, D. S. Determination of Residual Bromate in Baked Goods by Ion Chromatography with ICP-MS Detection. Paper presented at the 1994 Winter Conference on Plasma Spectrochemistry, San Diego, CA, January 10-15, 1994. (20) Heitkemper, D. T.; Kaine, L. A.; Jackson, D. S.; Wolnik, K. A. J. Chromatogr., A. 1994, 671, 101. (21) Heitkemper, D. T.; Kaine, L. A. Application of IC-ICPMS/AES in Foods and Dietary Supplements. Paper presented at the Twenty-Second Annual Conference of the Federation of Analytical Chemistry and Spectroscopy Societies, Cincinnati, OH, October 15-20, 1995. (22) Heumann, K. G.; Diemer, J.; Gallus, S.; Vogl, J. Accurate Elemental Speciation By Chromatographic Methods Coupled With Isotope Dilution Mass Spectrometry. Paper presented at the Twenty-Third Annual Conference of the Federation of Analytical Chemistry and Spectroscopy Societies, Kanasas City MO, September 29, 1996. (23) Diemer, J.; Heumann, K. G. Fresenius J. Anal. Chem. 1997, 357, 74. (24) Creed, J. Brockhoff, C. The determination of bromate in drinking water by Ion Chromatography ICPMS using Isotope Dilution. Paper presented at 45th ASMS Conference on Mass Spectrometry and Applied Topics, Palm Springs CA, June 1-5, 1997. (25) Creed, J.; Magnuson, M.; Brockhoff, C. Environ. Sci. Technol. 1997, 31, 2059.

used for the preparation of the RP cartridges (Dionex, Sunnyvale, CA) was Optima grade (Fisher). The synthetic chloride matrix was made from NaCl (J. T. Baker, Phillipsburg, NJ), the sulfate matrix was made from K2SO4 (Fisher) and the phosphate matrix was made from KH2PO4 (Fisher). The isotopically enriched sodium bromide 79 and 81 standards (Cambridge, Andover, MA) were 99.4 and 98.7% pure, respectively. The enriched bromate standards were produced by adjusting the pH of the enriched NaBr standard to 9 (with NaOH) and then ozonating the standard on a bench-scale ozonator (Griffin Technics, Lodi, NJ, model GTC 025) until the residual bromide was not detectable (∼1 ng/g) in the standard. All glassware was carefully cleaned to ensure that natural bromide was not introduced from the laboratory glassware into the contact chamber. All solutions that came in contact with the enriched material were analyzed for natural bromide and were found to contain less than detectable (∼1 ng/g) quantities of bromide. The naturally occurring 79/81 bromine ratio was monitored via a postcolumn injection of bromate, and the mass calibration was verified weekly. The data presented in this paper are not corrected for mass bias. The isotope dilution equation used has been reported by Taylor.26 Finally, the density of water is considered to be unity for dilution and standard preparation. Apparatus. The ICPMS was a PQ1 (Fisons, Beverly, MA) which was upgraded with a high-performance interface. The data were collected using the time-resolved or chromatographic software. The dwell time for each mass was set to 0.4 s. This time was chosen based on isotope ratio precision measurement using a flow injection analysis (FIA) peak. The chromatographic peaks were integrated off-line using a routine written in SAS (SAS Institute Inc., Cary, NC). This routine uses a quadratic fit to estimate the baseline and uses this baseline to integrate the chromatographic peaks. The plasma, auxiliary, and nebulizer flow rates were 14, 0.8, and 0.95 L/min, respectively. The ICP forward power was 1.3 kW and was not optimized for bromate analysis. The ion chromatograph consists of a Dionex gradient pump (model GPM2) and two six-way valves. The self-regenerating suppressor (Dionex) exchanges the sodium ion in the mobile phase for hydronium ion. This exchange eliminates the deposition of sodium on the ICPMS sampling cone and thereby provides improved long-term precision. The flow rate of the distilled, deionized water through the self-regenerating suppressor was 1 mL/min. The PA-100 guard (capacity 18 µequiv)/analytical (capacity 90 µequiv) columns have an alkyl quaternary amine functional group. The chromatographic flow rate was 1.0 mL/ min. All columns were purchased from Dionex. The samples were pretreated using a RP cartridge (Dionex) to remove the trisubstituted haloacetic acids.25 The cartridges were prepared and used according to manufacturer’s recommendations. The chromatographic setup has one unique feature. A 100-µL postcolumn standard injection loop is used to inject a bromate standard, which does not traverse the column. This standard synchronizes data collection from the mass spectrometer with the start of the ion chromatograph’s program. This postcolumn injection was used as a drift standard in the direct analysis mode and also used as an ongoing mass 79/81 ratio check in isotope (26) Montaser, A.; Golightly, G. W. Inductively Coupled Plasmas in Analytical Atomic Spectrometry; VCH Publishers: New York,1992; p 658.

Analytical Chemistry, Vol. 71, No. 3, February 1, 1999

723

Table 1. Precision and Recovery Data for Isotope Dilution Analysis as a Function of Bromate Concentrationa bromate std concn

fortified withb 79BrO3 av rec ( % RSD

fortified with3 81BrO3 av rec ( % RSD

50 10 1

101 ( 3.1 101 ( 5.4 102 ( 10.4

99 ( 2.0 103 ( 4.8 111 ( 14.6

direct analysis av rec ( % RSD 99 ( 3.2

a Each value was reported based on five replicate analyses. b The average determined mass ratios (79/81) for 50, 10, and 1 ng/g standards were 2.30, 2.34, and 2.30, respectively. c The average mass ratios for (81/79) for 50, 10, and 1 ng/g standards were 2.45, 2.40, and 2.32, respectively.

Figure 1. Retention times for an interference test mixture vs mM NH4NO3 in eluent. Retention times based on peak height.

dilution analysis mode. The sample is injected onto the column using a 580-µL loop. This sample loop volume is relatively large but is compatible with common anion concentrations in drinking water matrixes. The somewhat broader chromatographic peaks and loss of resolution induced by the larger sample loop volume is at least in part compensated for by selectivity of the detector which narrows the list of potential interferences and, therefore, the required chromatographic resolution. RESULTS AND DISCUSSION Enriched Na79BrO3 and Na81BrO3 Standards. The enriched standards were made utilizing a bench-scale ozonator as described in the Reagents section. The standards were then diluted to 400 ng/g and analyzed five times using the above chromatographic conditions. The purity for the 79 and 81 were 99.4 ( 0.07 (x ( 2σ) and 98.7 ( 0.01% (x ( 2σ), respectively. The resulting solutions did not contain detectable (∼1 ng/g) quantities of bromide. These values agree to within the instrumental precision of those reported by Cambridge (79Br, 99.4%; 81Br, 98.7%). Column and Mobile-Phase Considerations. The anionic character of brominated haloacetic acids at a near-neutral pH can cause chromatographic interferences with bromate when anionion chromatography is used as a separation scheme.25 The use of 5 mM HNO3 as a mobile phase resolves bromoacetic acid from bromate.25 However, the resulting NO3- solvent strength limits the tolerance for >100 ppm (2.8 mM) chloride concentrations because of peak broadening and matrix-induced elution effects.27 Drinking waters within the United States can contain chloride at concentrations greater than 100 ppm, and for this reason, a separation with greater matrix tolerance was required. Figure 1 investigates the addition of NH4NO3 to the mobile phase as a means of increasing the separations tolerance to drinking water anion matrixes while preserving the long-term instrument stability associated with a 5 mM HNO3 mobile phase and adequate resolution from known interferences. Figure 1 is a plot of retention times of a test mixture vs millimolar NH4NO3. The test mixture contains bromoacetic acid (which has retention characteristics similar to that of bromate), bromate, dibromoacetic acid (which is a late-eluting disubstituted haloacetic acid that is not removed by the RP cartridge), and bromide. Figure 1 indicates a 5 mM (27) Novic, M.; Divjak, B.; Pihlar, B.; Hudnik, V. J. Chromatogr., A 1996, 739, 35.

724 Analytical Chemistry, Vol. 71, No. 3, February 1, 1999

HNO3 mobile phase in combination with a 20-25 mM NH4NO3 produces an adequate separation of bromate from bromoacetic acid while minimizing the overall analysis time. The 5 mM HNO3 + 25 mM NH4NO3 was chosen for all subsequent work in an attempt to minimize analysis time. Recovery and Precision Comparison for Direct and Isotope Dilution Analysis. The ability to quantitate using isotope dilution is at least partially dependent on the precision of the determined ratio and, with a chromatographic-based technique, the ability to integrate the associated peaks. The two isotopes of bromine, masses 79 and 81, have polyatomic backgrounds induced by 40Ar38ArH and 40Ar40ArH ions. These backgrounds are, in effect, subtracted when the chromatographic baseline is determined. The assumption is that the production of the polyatomic is constant over the retention window for bromate. The difference in the natural abundances (40Ar ) 99.6%, 38Ar ) 0.337%) produces a larger background and, in turn, a larger signal variation on mass 81. Therefore, mass 79 is used for quantitation in the direct analysis mode. Table 1 compares the precision and recovery data at three different concentrations for isotope dilution analysis and compares the precision of isotope dilution to direct analysis at 10 ng/g. The precision and recovery is determined by using a known standard containing 1, 10, and 50 ng/g bromate. The standard concentration is used to calculate the required isotopic addition for isotope dilution analysis. The precision of isotope dilution analysis is improved by utilizing molar ratios near unity.28 To approximate this ideal ratio, the enriched isotopic additions used in Table 1 were made so the resulting mass 79/81 ratio was between 2.3 and 2.5. The relative standard deviations (RSDs) at 50 ng/g reported in Table 1 are approximately 2-3% for samples fortified with either isotope. RSDs of 2-3% represent the precision limit for IC chromatographic isotope dilution analysis using our experimental setup. This precision is relatively poor in comparison to precisions that are commonly associated with isotope ratio measurements via ICPMS. The added imprecision may be induced by the transient nature of the signal vs conventional nebulization or the spectral backgrounds associated with bromine detection via ICPMS. In Table 1, a comparison of precision is made between direct analysis and isotope dilution analysis at 10 ng/g. This concentration was selected because this is the “expected” maximum contaminant level (MCL) for bromate. The isotope dilution (28) Garbarino, J. R.; Taylor, H. E. Anal. Chem. 1987, 59, 1568.

Table 2. Precision and Recovery Data for Bromate in Sulfate, Phosphate, and Chloride Matrixesa isotope dilution

matrix 300 ppm chloride 300 ppm sulfate 100 ppm phosphate a

10 ng/g BrO3- + 10 ng/g BrO3- + direct analysis 7.7 ng/g 81BrO3- 6.4 ng/g 79BrO3- 25 ng/g BrO3av rec ( % RSD av rec ( % RSD av rec ( % RSD 100 ( 2.2 97 ( 2.1 98 ( 1.8

105 ( 8.6 104 ( 1.5 97 ( 3.8

101 ( 2.8 112 ( 2.5 98 ( 1.9

All values were reported based on five replicate anlayses.

analysis and direct analysis produce RSDs of approximately 5 and 3%, respectively. These RSDs are certainly acceptable from a required precision for bromate monitoring in drinking water at 10 ng/g, but the precision commonly associated with isotope dilution analysis is usually superior in comparison to a direct analysis. (i.e., the precision in determining a ratio is superior to determining a sample concentration (without matrix effects) from a calibration curve). Diemer et al.23 reported relative standard deviations of 13% for a bromate concentration of 3 µg/L using isotope dilution IC-ICPMS with 700 and 15 000 counts/s background signals on masses 79 and 81, respectively. One possible explanation for the observed comparable precision (direct vs isotope dilution) is that the imprecision in monitoring and integrating mass 81 (required in isotope dilution analysis and not required in direct analysis) approximates the imprecision induced by normal instrumental drift over a 40-50min analysis period (required for direct analysis and not required for isotope dilution). Therefore, direct analysis of bromate compares favorably to isotope dilution analysis in terms of recovery and precision if instrument drift is held to 2-4% over the direct analysis period. The mobile phase of 5 mM HNO3 + 25 mM NH4NO3, chosen for these experiments, helps to minimize instrumental drift caused by mobile-phase deposition on the ICPMS sampling and skimmer cones. Alternatively, increased ratio precision may be obtained by minimizing the polyatomic 40Ar40ArH ion and, in turn, the variation on mass 81. Finally, in Table 1, the RSDs for a 1 ng/g bromate solution have increased to 1015% because of the increased variation typical of concentrations approaching the detection limit. The detection limit for the direct analysis of bromate was determined to be 0.3 ng/g based on seven replicate analyses of a low-level standard.29 Overall, Table 1 indicates the precision for isotope dilution analysis is comparable to the direct analysis given the experimental conditions utilized. The precision at the concentrations reported for either analysis mode is acceptable for drinking water analysis of bromate, while the isotope dilution analysis provides a means of validating the direct analysis approach. In addition, the use of isotope dilution analysis allows for an on-line check of potential analyte losses during IC preconcentration applications. Matrix Anion Tolerance. Two of the limitations of using a 5 mM HNO3 mobile phase are matrix tolerance and peak shape degradation.25 This limitation is at least partially attributable to the anionic strength of the mobile phase.27 Table 2 reports the precision and recovery data for bromate in three common drinking (29) Glaser, J. A.; Foerst, D. L.; McKee, G. D.; Quave, S. A.; Budde, W. L. Environ. Sci. Technol., 1981, 15, 1426.

Figure 2. Polyatomic ions produced on mass 81 from 300 ppm sulfate matrix.

water matrixes: chloride, sulfate, and phosphate. The chloride, sulfate, and phosphate concentrations used in Table 2 represent a rough upper limit for these matrix anions in drinking water supplies within the United States. The recovery of 10 ng/g bromate via isotope dilution in 300 ppm chloride, 300 ppm sulfate, and 100 ppm phosphate are within 3% of 100%. The RSDs for isotope dilution analysis in these three matrixes are less than 3%. These results compare favorably with the direct analyses of bromate at 25 ng/g except in the presence of sulfate. The recovery of 10 ng/g bromate via direct analysis in 300 ppm sulfate produces an average recovery of 112% with a RSD of 2.5%. The reason for this bias in a sulfate matrix is not currently understood. The recoveries obtained in the chloride and phosphate matrix indicate the column and detection system are capable of monitoring bromate in most drinking water matrixes found in the United States. The 300 ppm sulfate and 100 ppm phosphate matrixes produce polyatomic species as they elute from the ion chromatograph. Therefore, these species must be chromatographically resolved from bromate. Figure 2 is a chromatogram of 10 ng/g BrO3- + 7.7 ng/g81BrO3- in a 300 ppm sulfate matrix. The sulfate elutes as a mixture of HSO4- and SO42- (pKa ) 1.9) at an eluent pH of 2.7-2.8 producing a polyatomic ion on mass 81 tentatively identified as HSO3+. This polyatomic ion has been identified in sulfuric acid matrixes by Tan and Horlick.30 This anion does not influence the detection of bromate via direct analysis because of its chromatographic resolution, and mass 79 is used for direct analysis quantitation. The recovery via isotope dilution analysis in a 300 ppm sulfate matrix in Table 2 indicates this polyatomic does not influence the integration on mass 81. Figure 3 is a chromatogram of 10 ng/g BrO3- + 7.7 ng/g 81BrO3- in a 100 ppm phosphate matrix. The 100 ppm phosphate matrix generates two polyatomic anions as it elutes from the column. Unlike sulfate, phosphate elutes prior to bromate. The phosphate elutes as a mix of H2PO4- and H3PO4 (pKa ) 2.1) at an eluent pH of 2.7-2.8 producing a polyatomic on masses 81 and 79 tentatively identified as H2PO3+ and PO3+, respectively. The production of these two polyatomics does not produce a recovery bias at 100 ppm phosphate as evident in Table 2, while a higher concentration may (30) Tan, S. H.; Horlick, G. Appl. Spectrosc. 1986, 40, 445.

Analytical Chemistry, Vol. 71, No. 3, February 1, 1999

725

Table 3. Bromate Analysis via IC-ICPMS of Finished Ozonated Drinking Watersa

Figure 3. Polyatomic ions produced on masses 79 and 81 from 100 ppm phosphate matrix.

start to degrade the baseline determination for bromate. The polyatomic ions observed for sulfate and phosphate are unlike the brominated haloacetic acids in that the polyatomic ions will produce incorrect Br 79/81 ratios while the brominated haloacetic acids produce 79/81 ratios near unity. Sulfate retention time is close to dibromoacetic acid, and the 79/81 ratio can be used to verify the presence of sulfate. Similarly, phosphate retention time is similar to bromoacetic acid, and the ratio can be used to confirm the presence of phosphate. Chloride elutes immediately after bromate and broadens the resulting bromate peak. This can lead to software integration difficulty when trace bromate (300 ppm) chloride matrixes. However, the use of 5 mM HNO3 + 25 mM NH4NO3 provided adequate resolution of bromate from the known potential interferences (bromide containing and polyatomic) while providing increased (relative to 5 mM HNO325) matrix tolerance to peak shape degradation in high chloride matrixes. Precision and Recovery in Finished Ozonated Drinking Waters. Table 3 contains precision and recovery data for bromate from 10 independent drinking water supplies from across the United States. All the waters in Table 3 are finished drinking waters, which means they were sampled just prior to entering the distribution system. The 10 waters were analyzed by direct analysis and by isotope dilution analysis. The required isotopic addition was calculated based on the concentration determined from a single direct analysis, and the resulting determined ratio ranged from 2.2 to 4.1. The determined bromate concentrations for isotope dilution and direct analysis are in good agreement with

726 Analytical Chemistry, Vol. 71, No. 3, February 1, 1999

determination by direct analysis

determination by isotope dilution

drinking water sample

concb (ng/g)

% RSD

concb (ng/g)

% RSD

1 2 3 4 5 6 7 8 9 10

22.2 3.0 10.1 2.7 1.3 0.8 4.5 3.7 0.6 1.5

5.5 6.4 3.6 5.1 10.6 15.5 5.1 5.0 23.1 25.9

23.0 3.0 10.1 2.7 1.0 0.6 4.5 3.4 1.0 1.3

2.4 5.4 6.1 7.2 7.7 29.9 8.9 7.3 19.0 19.1

analyses were based on n ) 5. b Samples were fortified with The determined ratio after the isotopic addition ranged from 2.1 to 4.1.

a All 81BrO-

3.

the largest percent difference occurring in water 9. This concentration is ∼3 times the detection limit. The RSDs for both techniques are 3-6% for bromate concentrations above 10 ng/g, while the RSDs are 5-10% for bromate concentrations from 2 to 10 ng/g. Known potential interferences were identified in water 4, bromochloroacetic acid, and dibromoacetic acid, water 7 contained phosphate-producing H2PO3+, and water 8 contained sulfate-producing HSO3+. The results in Table 3 indicate comparable analytical capability for isotope dilution vs direct analysis of bromate via IC-ICPMS. ACKNOWLEDGMENT The authors acknowledge Richard Miltner from the National Risk Management Research Laboratory in Cincinnati, OH, and Hiba Shukairy from the Technical Support Center in Cincinnati, OH, for their technical assistance in producing the enriched bromate standards. The authors also acknowledge Doug Heitkemper and Lisa Kaine from the FDA Forensic Chemistry Center in Cincinnati, OH, for their helpful discussions on matrix effects encountered in the separation of bromate in extreme chloride matrixes. Received for review June 16, 1998. Accepted October 15, 1998. AC980663N