Development of Isotope Dilution Cold Vapor Inductively Coupled

Development of Isotope Dilution Cold Vapor Inductively Coupled Plasma Mass Spectrometry and Its Application to the Certification of Mercury in NIST St...
76 downloads 13 Views 125KB Size
Anal. Chem. 2001, 73, 2190-2199

Development of Isotope Dilution Cold Vapor Inductively Coupled Plasma Mass Spectrometry and Its Application to the Certification of Mercury in NIST Standard Reference Materials S. J. Christopher*

National Institute of Standards and Technology-Charleston Laboratory, 219 Fort Johnson Road, Charleston, South Carolina 29412 S. E. Long, M. S. Rearick, and J. D. Fassett

National Institute of Standards and Technology, 100 Bureau Drive, Gaithersburg, Maryland 20899-8391

An isotope dilution cold vapor inductively coupled plasma mass spectrometry (ID-CV-ICPMS) method featuring gaseous introduction of mercury via tin chloride reduction has been developed and applied to the quantification and certification of mercury in various NIST standard reference materials: SRM 966 Toxic Metals in Bovine Blood (30 ng‚mL-1); SRM 1641d Mercury in Water (1.6 µg‚mL-1); and SRM 1946 Lake Superior Fish Tissue (436 ng‚g-1). Complementary mercury data were generated for SRMs and NIST quality control standards using cold vapor atomic absorption spectroscopy (CVAAS). Certification results for the determination of mercury in SRM 1641d using two independent methods (ID-CV-ICPMS and CVAAS) showed a degree of agreement of 0.3% between the methods. Gaseous introduction of mercury into the ICPMS resulted in a single isotope sensitivity of 2 × 106 counts‚s-1/ng‚g-1 for 201Hg and significantly reduced the memory and washout effects traditionally encountered in solution nebulization ICPMS. Figures of merit for isotope ratio accuracy and precision were evaluated at dwell times of 10, 20, 40, 80, and 160 ms using SRM 3133 Mercury Spectrometric Solution. The optimum dwell time of 80 ms yielded a measured 201Hg/202Hg isotope ratio within 0.13% of the theoretical natural value and a measurement precision of 0.34%, on the basis of three replicate injections of SRM 3133. Concern about the health effects of mercury in the environment is rapidly increasing. The ability to determine accurately trace amounts of mercury in a wide range of sample types is required to assess these health effects and to identify global, regional, and point sources that release mercury into the atmosphere and natural waters. Anthropogenic inputs of mercury from combustion sources, such as fossil fuel burning and medical waste incineration, result in the accumulation of mercury in aquatic food webs. It is believed that fish and shellfish consumption is the principal route for human exposure to mercury.1 2190 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

New regulatory initiatives for mercury are anticipated as a result of its toxicity and ubiquitous nature in the environment. These initiatives may force environmental controls on mercury emission that ultimately could have significant economic impact. It is important that new and evolving risk assessments for trace amounts of mercury in the environment are based on solid scientific evidence; thus, critical reference materials (RMs) are needed to benchmark the accuracy of analytical methods that may be applied. A sensitive, reliable, and accurate analytical method, such as isotope dilution mass spectrometry (IDMS), ensures the accuracy of the RM certification process. Various spectroscopic, electrochemical, and radiochemical methods exist for the determination of trace amounts of mercury. Clevenger, Smith, and Winefordner published a 1997 review article that surveys the instrumental techniques used for the quantitation of mercury.2 The atomic spectroscopic methods used for the determination of mercury include atomic absorption spectroscopy,3-10 atomic fluorescence spectroscopy,11-17 and inductively (1) Mercury Report to Congress (EPA 452/R-97-0003), December, 1997. (2) Clevenger, W. L.; Smith, B. W.; Winefordner, J. D. Crit. Rev. Anal. Chem. 1997, 27, 1-26. (3) Costley, C. T.; Mossop, K. F.; Dean, J. R.; Garden, L. M.; Marshall, J.; Carroll, J. Anal. Chim. Acta 2000, 405, 179-183. (4) Neto, J. A. G.; Zara, L. F.; Rocha, J. C.; Santos, A.; Dakuzaku, C. S.; Nobrega, J. A. Talanta 2000, 51, 587-594. (5) Tao, G. H.; Willie, S. E.; Sturgeon, R. E. J. Anal. Atom. Spectrom. 1999, 14, 1929-1931. (6) Hardisson, A.; Padron, A. G.; de Bonis, A.; Sierra, A. Atom. Spectrosc. 1999, 20, 191-193. (7) de Mirabo, F. M. B.; Thomas, A. C.; Rubi, E.; Forteza, R.; Cerda, V. Anal. Chim. Acta 1997, 355, 203-210. (8) Sakamoto, H.; Taniyama, J.; Yonehara, N. Anal. Sci. 1997, 13, 771-775. (9) Ombaba, J. M. Microchem. Journal 1996, 53, 195-200. (10) Nixon, D. E.; Mussmann, G. V.; Moyer, T. P. J. Anal. Toxicol. 1996, 20, 17-22. (11) Peart, D. B.; Antweiler, R. C.; Taylor, H. E.; Roth, D. A.; Brinton, T. I. Analyst 1998, 123, 455-476. (12) Corns, W. T.; Stockwell, P. B.; Jameel, M. Analyst 1994, 119, 2481-2484. (13) Bramanti, E.; D’Ulivo, A.; Lampugnani, L.; Zamboni, R.; Raspi, G. Anal. Biochem. 1999, 274, 163-173. (14) Jones, R. D.; Jacobson, M. E.; Jaffe, R.; Westhomas, J.; Arfstrom, C.; Alli, A. Water, Air, Soil Pollut. 1995, 80, 1285-1294. (15) Hall, G. E. M.; Pelchat, P. Water, Air, Soil Pollut. 1999, 111, 287-295. 10.1021/ac0013002 CCC: $20.00

© 2001 American Chemical Society Published on Web 04/10/2001

coupled plasma mass spectrometry (ICPMS).18-31 The ability to monitor isotope ratios using ICPMS advantageously allows for the high accuracy technique of isotope dilution to be implemented.32 Concerning mercury analysis using traditional (i.e., aqueous sample introduction) ICPMS, isotope dilution inductively coupled plasma mass spectrometry (ID-ICPMS) methods are often implemented because they address many of the low-recovery problems often encountered for mercury when alternate calibration schemes are used.22,24 However, the ID-ICPMS technique employed for mercury in solution can still suffer from long mercury washout times, reduced sensitivity, memory effects, and isotope ratio imprecision. These limitations often plague accurate and efficient determinations of mercury, especially for samples containing naturally low (parts per billion) levels of the analyte. Gaseous injection of mercury via hydride or cold vapor generation directly into an inductively coupled plasma19,24,28 is often implemented in order to mitigate some of the problems associated with solution nebulization; that is, to reduce memory effects and to increase sensitivity and element selectivity in the absence of sample matrix ions. Introduction of mercury into an ICPMS by electrothermal heating is also used to address these issues.20,33 In recent years, Jiang and co-workers published a series of articles dealing with the determination of mercury by ICPMS using vapor generation or electrothermal vaporization (ETV) methodologies.34-42 Often, the ETV determinations were coupled with the isotope dilution technique.35,37-39,41 The ID-ETV-ICPMS method for de(16) Rea, A. W.; Keeler, G. J. Biogeochemistry 1998, 40, 115-123. (17) Armstrong, H. L.; Corns, W. T.; Stockwell, P. B.; O’Connor, G.; Ebdon, L.; Evans, E. H. Anal. Chim. Acta 1999, 390, 245-253. (18) Yoshinaga, J.; Morita, M. J. Anal. Atom. Spectrom. 1997, 12, 417-420. (19) Brown, R.; Gray, D. J.; Tye, D. Water, Air, Soil Pollut. 1995, 80, 12371245. (20) Smith, R. G. Anal. Chem. 1993, 65, 2485-2488. (21) Campbell, M. J.; Vermeir, G.; Dams, R.; Quevauviller, P. J. Anal. Atom. Spectrom. 1992, 7, 617-621. (22) Jian, L.; Goessler, W.; Irgolic, K. J. Fresenius’ J. Anal. Chem. 2000, 366, 48-53. (23) Nixon, D. E.; Burritt, M. F.; Moyer, T. P. Spectrochim. Acta, Part B 1999, 54, 1141-1153. (24) Knight, R.; Haswell, S. J.; Lindow, S. W.; Batty, J. J. Anal. Atom. Spectrom. 1999, 14, 127-129. (25) Allibone, J.; Fatemian, E.; Walker, P. J. J. Anal. Atom. Spectrom. 1999, 14, 235-239. (26) Choi, M. H.; Cech, J. J. Environ. Toxicol. Chem. 1998, 17, 1979-1981. (27) Moreton, J. A.; Delves, H. T. J. Anal. Atom. Spectrom. 1998, 13, 659-665. (28) Karunasagar, D.; Arunachalam, J.; Gangadharan, S. J. Anal. Atom. Spectrom. 1998, 13, 679-682. (29) Woller, A.; Garraud, H.; Martin, F.; Donard, O. F. X.; Fodor, P. J. Anal. Atom. Spectrom. 1997, 12, 53-56. (30) Stroh, A.; Vollkopf, U. J. Anal. Atom. Spectrom. 1993, 8, 35-40. (31) Haraldsson, C.; Westerlund, S.; Ohman, P. Anal. Chim. Acta 1989, 221, 77-84. (32) Montaser, A. Inductively Coupled Plasma Mass Spectrometry; Wiley-VCH: New York, 1998; p 699. (33) Snell, J. P.; Bjorn, E.; Frech, W. J. Anal. Atom. Spectrom. 2000, 15, 397402. (34) Wan, C. C.; Chen, C. S.; Jiang, S. J. J. Anal. Atom. Spectrom. 1997, 12, 683-687. (35) Wei, M. T.; Jiang, S. J. J. Chin. Chem. Soc. 1999, 46, 871-878. (36) Chen, C. S.; Jiang, S. J. Spectrochim. Acta, Part B 1996, 51, 1813-1821. (37) Liao, H. C.; Jiang, S. J. J. Anal. Atom. Spectrom. 1999, 14, 1583-1588. (38) Liu, H. W.; Jiang, S. J.; Liu, S. H. Spectrochim. Acta, Part B 1999, 54, 13671375. (39) Liao, H. C.; Jiang, S. J. Spectrochim. Acta, Part B 1999, 54, 1233-1242. (40) Lee, K. H.; Jiang, S. J.; Liu, H. W. J. Anal. Atom. Spectrom. 1998, 13, 12271231. (41) Chen, S. F.; Jiang, S. J. J. Anal. Atom. Spectrom. 1998, 13, 1113-1117. (42) Chen, S. F.; Jiang, S. J. J. Anal. Atom. Spectrom. 1998, 13, 673-677.

termination of mercury is very promising, although the cost and complexity of ETV equipment and the need for matrix modifiers may deter some end-users from pursuing this method. Mercury vapor generation procedures based on simple equipment and reduction chemistry as described here offer a powerful means to address the accuracy, sensitivity, and washout limitations of solution nebulization ICPMS. The results reported here for determination of mercury in SRM 966 Toxic Metals in Bovine Blood; SRM 1641d Mercury in Water; SRM 1946 Lake Superior Fish Tissue; and two NIST quality control materials prepared from fresh-frozen livers of pilot and beluga whales demonstrate the first application of isotope dilution cold vapor inductively coupled plasma mass spectrometry (ID-CV-ICPMS) to the certification of mercury in reference materials. EXPERIMENTAL SECTION Reagents. Nitric acid was either purified at NIST with a subboiling distillation apparatus or purchased (optima grade) from Fisher Scientific (Suwanee, GA). Trace metal certified hydrogen peroxide 30% (w/w) was purchased from JT Baker (Phillipsburg, NJ). For the isotope dilution experiments, a 100 ng‚g-1 mercury spike solution in 10% (w/w) HNO3 was created from 201Hgenriched solid mercuric oxide that was purchased from Oak Ridge National Laboratory (Oak Ridge, TN). SRM 3133 mercury spectrometric solution was obtained from NIST (Gaithersburg, MD) and used as a spike for the standard addition experiments. All of the sample and standard solutions were diluted with highquality water that was obtained from either a NIST two-stage quartz distillation apparatus or a Millipore deionization station that was capable of producing 18 MΩ resistivity water. Tin chloride, hydrochloric acid, and potassium dichromate were purchased from JT Baker. The tin chloride reductant solution, 10% (w/w) SnCl2 in 7% (w/w) HCl, was vigorously agitated to liberate any background elemental mercury prior to use. Sample Preparation. SRM 966 Toxic Metals in Bovine Blood. The base material for this SRM was prepared at a USDA-licensed facility under the direction of the Centers for Disease Control and Prevention (CDC). Cows were dosed with lead nitrate and then bled in units of 500 mL. The blood was then mixed with EDTA to prevent microclotting and stored at 4 °C. Multiple pools of blood were prepared. One of the pools was spiked with methyl mercury (MeHg as iodide) and inorganic mercury. Each sample vial of SRM 966 contains a 2 to 3 mL aliquot of bovine whole blood. Units of SRM 966 are issued in paired sets containing baseline (natural) levels of elements, including mercury, designated level I samples, and elevated (spiked) levels of elements, designated level II samples. The concentration of total mercury in the level II samples is approximately 30 ng‚mL-1. Vials of SRM 966 were stored frozen at - 20 °C and were equilibrated to room temperature for 2 h prior to processing. A Milestone (Monroe, CT) microwave system-model MLS 1200 was used to effect sample decomposition using the parameters outlined in Table 1. For the microwave dissolution procedure, 2-g samples of SRM 966 were weighed directly into microwave cells in a decomposition medium of 5 mL of concentrated HNO3 + 1 mL of 30% (w/w) H2O2. Various procedural blanks and a control standard, SRM 2976 Mussel Tissue (trace elements and methylmercury), were digested along with the SRM 966 samples. The resulting digests were transferred into preweighed 45-mL polyAnalytical Chemistry, Vol. 73, No. 10, May 15, 2001

2191

Table 1. Microwave Digestion Routines Used for Various Samplesa sample

acid medium

SRM 966 Toxic Metals in Bovine Blood

5 mL HNO3, 1 mL H2O2

SRM 1946 Lake Superior Fish Tissue

5 mL HNO3, 2 mL HClO4

beluga, pilot whale liver homogenates

5 mL HNO3

time min

power W

7

250

6 4 2 4

400 550 250 600

5

250

5 20 5 10

400 500 250 500

0-5 20

0-100 ramp 800

microwave equipment Milestone MLS 1200

Milestone MLS 1200

Perkin-Elmer Multiwave

a No microwave digestion routine used for SRM1641d Mercury in Water.

ethylene bottles, and the solution in each bottle was diluted to about 25 g (group 1 solutions). To stabilize mercury, a 3% (w/w) HNO3 solution containing 0.05% (w/w) K2Cr2O7 was used as the diluent. Aliquots (12 g) of the group 1 solutions were subsampled and appropriately spiked with about 2 g of a 50 ng‚g-1 mercury spike solution for the purpose of quantification of total mercury by cold vapor atomic absorption spectroscopy (CVAAS) and the method of standard additions. Of the 14 samples, 4 samples were spiked with an enriched isotopic 201Hg standard that served a dual purpose: it acted as a standard addition spike for the CVAAS measurements as well as the isotopic spike for the ID-CV-ICPMS measurements. Spiking the samples in this manner allowed for a direct comparison of instrumental bias. Ideally, one would add the isotopic spike to the sample prior to microwave digestion to ensure complete spike equilibration; however, because a direct comparison between the CVAAS and ID-CV-ICPMS data was desired, the necessity to split the samples for the CVAAS standard addition experiment forced the isotopic spike to be added after microwave digestion had occurred. SRM 1641d Mercury in Water. Microwave dissolution procedures were not used for water samples of SRM 1641d and the SRM 1641c control; however, the SRM 1641 samples were similarly diluted and stabilized as described above. In addition, two independent sample populations (10 each) were appropriately spiked and analyzed on separate days by either standard additions CVAAS (SRM 3133 spike) or ID-CV-ICPMS (201Hg isotopic spike). This allowed for the certification of SRM 1641d using two independent methods. SRM 1946 Lake Superior Fish Tissue. Vials of SRM 1946 were stored frozen at - 80 °C and temporarily stored in a thermally insulated container on dry ice while weighing. Approximately 0.30.5-g samples of SRM 1946 were weighed directly into microwave cells with the aid of a frozen Teflon spatula. Along with the SRM 1946 samples, a control sample of SRM 2976 Mussel Tissue (trace elements and methylmercury), and four procedural blanks were spiked with an enriched 201Hg isotopic standard prior to microwave digestion. The same Milestone microwave system was used to 2192

Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

effect sample decomposition using the parameters outlined in Table 1, in a manner similar to the SRM 966 samples. The microwave decomposition medium consisted of 4 mL of concentrated HNO3 and 2 mL of concentrated HClO4. Independent samples were not prepared for CVAAS. NIST Whale Liver Quality Control Samples. For the purposes of a quality assurance exercise, fresh-frozen samples of pilot and beluga whale liver homogenates were prepared for analysis by ID-CV-ICPMS, as well as standard additions CVAAS. The microwave dissolution procedure involved weighing 0.5-0.7 g samples of the whale liver homogenates directly into the microwave cells, along with four procedural blanks in a decomposition medium of 5 mL of concentrated HNO3. All of the samples were then digested in a Perkin-Elmer (Boston, MA) Multiwave microwave oven using the power/time profile tabulated in Table 1. For determination of mercury, the resulting digests were quantitatively transferred into preweighed 60-mL polyethylene bottles, and the solution in each bottle was diluted to about 60 g (group 1 solutions). Because of the large concentration of mercury in the samples (∼28 mg‚kg-1 and 42 mg‚kg-1, respectively, for pilot and beluga whales), additional stages of serial dilution were needed prior to analysis. Separate 1-g aliquots were sampled from the group 1 solutions and spiked with either an enriched 201Hg isotopic standard for ID-CV-ICPMS or an SRM 3133 spike for standard additions CVAAS analysis. The whale liver samples designated for ID-CVICPMS analysis were run in conjunction with the SRM 1946 samples above. Safety Considerations. For safety reasons, it should be noted that perchloric acid poses a fire/explosion hazard when handled and should only be used by qualified personnel with appropriate equipment. In addition, methylmercury (present in SRM 966) is a highly toxic substance that can readily pass through personal protective equipment and be absorbed through the skin and mucous membranes. Instrumentation. All CVAAS determinations were performed on a CETAC (Omaha, NE) model M6000 mercury analyzer using a reductant solution of 10% (w/w) SnCl2 in 7% (w/w) HCl. Typical experimental parameters include Ar flow rates of ∼150 mL‚min-1, mercury lamp currents of 6-8 mA, sample sip times of 20-30 s, and sample rinse times of 90 s. For the ICPMS measurements, a VG Elemental (Winsford, U.K.) PQ2 ICPMS was used in the timeresolved-analysis (TRA), peak-jumping mode in order to capture the transient 201Hg and 202Hg signals after elemental mercury was generated in the AA unit’s gas-liquid separator (GLS) reduction cell. Typical ICP parameters were used: 1350 W forward power, cooling gas flow rates of 13.5 L‚min-1, and injector flow rates of 0.81 L‚min-1. For all of the samples, reverse isotope dilution was used to calibrate the 201Hg isotopic spike solutions with a NIST primary standard (SRM 3133) and all of the data were corrected for detector dead time. RESULTS AND DISCUSSION Method Development. The use of argon carrier gas and low flow rates (150 mL‚min-1) in the CVAAS system allowed for a simple interfacing to the ICPMS instrument. Several ID-CV-ICPMS interfaces were tried before an optimized configuration was found. Figure 1 and the following three paragraphs describe the chronological evolution of the ICPMS interface. The Scott-type

Figure 1. Chronological evolution of the ICPMS interface. (a) Meinhard nebulizer interface, (b) plastic tee interface, (c) stand-alone interface.

double-pass spray chamber was removed from the system, regardless of the configuration. Initially, a Meinhard nebulizer was used to interface the AA unit’s waste carrier gas stream output to the ICPMS torch (Figure 1a). The waste gas stream was introduced into the central channel of the Meinhard nebulizer at a flow rate of 150 mL‚min-1 and combined with an injector gas flow rate of 0.81 L‚min-1. This arrangement facilitated injection of the Hg0 vapor along the ICP torch axis after an atomic absorbance measurement was collected. The AA unit’s autosampler (CETAC ASX-500) controlled sample uptake (∼6 mL/min), and a separate peristaltic pump channel controlled uptake of the reductant solution. All liquid and gaseous flows to the gas-liquid separator were controlled using software and hardware from the AA unit. This configuration was used to collect ICPMS data for the four isotopically spiked samples of SRM 966.

Several changes were made to the original interface in an attempt to improve the resultant isotope-time profiles. Eventually, the absorbance cell on the AA unit was bypassed to reduce the length of the mercury transfer tubing, and a plastic tee piece was substituted for the Meinhard nebulizer to decrease potential turbulence at the transfer line/ICP torch junction (Figure 1b). The peristaltic pump/autosampler arrangement and flow control hardware remained unchanged. This configuration was used to collect ICPMS data for the SRM 1641d samples. Finally, the rest of the samples were analyzed using an extremely simple arrangement that yielded the flattest transient isotope-time profiles (Figure 1c). The gas-liquid separator was altogether removed from the AA unit and positioned as near as practical to the ICPMS torch. The liquid sample flow rate was reduced to ∼2 mL/min, because the diameter of the sample uptake tubing was ∼10× smaller than that of the autosampler Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

2193

Figure 2. Typical isotope-time profile and 201Hg/202Hg ratios recorded for a spiked sample of SRM 966 using the Meinhard nebulizer interface.

probe tubing that was used in the two previous configurations (Figure 1a,b). A short 18-cm piece of 1.6-mm-diam Viton tubing served as the transfer line from the gas-liquid separator to the same plastic tee piece used in the second generation interface (Figure 1b). In addition, a separate mass-flow controller was used to control the flow rate of the Ar carrier stream through the gasliquid separator to eliminate the previous reliance on the AA unit’s flow-control software/hardware. Depending on the interface configuration, a sample timing delay from 10 to 25 s was set up to catch the transient 201Hg and 202Hg isotope-time profiles after elemental mercury was formed in the gas-liquid separator. This effectively allowed for each mercury transient signal to be bracketed by a flat background region representing when a 2% (w/w) HNO3 blank solution (not a sample) was present in the gas-liquid separator. This background offset was used to correct the signal for each mercury isotope measured, prior to calculating 201Hg/202Hg isotope ratios for the analytical samples. Isotope-Time Profiles and Analytical Figures of Merit for Different Interfaces. Meinhard Nebulizer Interface. It was during the certification of SRM 966 by CVAAS that the cold vapor technique was coupled with ID-ICPMS and evaluated utilizing the four isotopically spiked samples of SRM 966. The unspiked and 201Hg spiked samples were first analyzed by CVAAS, and then to facilitate analysis by ID-CV-ICPMS, the mercury waste carrier gas stream was introduced directly into the central channel of the ICPMS torch using the Meinhard nebulizer configuration described in the previous section. The standard additions CVAAS experiment required the analysis of both unspiked and spiked samples of SRM 966, but the ICPMS acquisition was only initiated to collect transient signals for spiked samples of SRM 966. Once the sample vapor reached the ICPMS, the 201Hg and 202Hg isotopes were monitored in TRA mode to recover the individual ion count rates. A dwell time of 40 ms was used for each isotope of mercury. Figure 2 shows a typical isotope-time profile recorded for the 2194 Analytical Chemistry, Vol. 73, No. 10, May 15, 2001

201Hg

and 202Hg isotopes, for an isotopically spiked sample of SRM 966 (vial 76). The transient profiles span a region of approximately 40 s and are nonuniform, likely owing to turbulence at the plasma torch injector interface or air entrainment in the autosampler’s uptake tubing. Plotting the background-corrected 201Hg/202Hg isotope ratios on the same time axis allows one to easily observe their variability across the transient profile. The 137 data points housed within the rectangle in Figure 2 at 25-38 s yield a 201Hg/202Hg ratio of 8.01 ( 0.28 (RSD ) 3.5%). Table 2 lists a summary of the CVAAS results for mercury in the level II samples of SRM 966 as well as the amount of mercury in the SRM 2976 control sample (derived from the mean of n ) 3 samples). Also included are the four ID-CV-ICPMS values for comparison purposes. The certified concentration of mercury in the level II SRM 966 samples is (31.5 ( 1.1) ng‚mL-1, a value that is in very good agreement with the concentration of 30 ng‚mL-1 targeted by the CDC. The mercury concentration results were corrected for density using a value of d ) 1.054 g‚mL-1. The expanded uncertainty was determined according to ISO guidelines.43 The uncertainty budget consisted of both type A and type B errors. Standard uncertainties were computed by combining sample-to-sample and blank measurement errors as type A errors, and dilution, spike calibration, and weighing errors as type B errors. Expanded uncertainties were computed by combining type A and type B errors in quadrature and multiplying by a coverage factor derived from the overall effective degrees of freedom and the Student’s t table. Percent difference results for samples 33, 74, 76, and 166 (Table 2) indicate that the ID-CV-ICPMS-derived mercury values are, on average, 1.4% lower than their corresponding CVAAS counterparts. A paired t-test (p ) 0.034) confirmed a small bias between the sample means. The source of this bias was not investigated, because modifications to the interface were anticipated. These (43) A Guide to the Expression of Uncertainty in Measurement; ISBN 92-67-101889, 1st ed.; ISO: Geneva, Switzerland, 1993.

Table 2. Summary Statistics Computed for Mercury in Samples of SRM 966 Using ID-CV-ICPMS and CVAAS Methods vial number

CVAAS data [Hg] ng‚mL-1

95 157 135 185 15 41 215 71 59 5 33 74 76 166 recommended value std dev % RSD expanded uncertainty 95%

29.43 29.53 33.68 29.31 31.80 30.66 29.69 33.81 33.07 33.59 32.92 29.79 29.99 33.94 31.52 1.904 6.04 1.1

ID-CV-ICPMS data [Hg] ng‚mL-1

Control Dataa CVAAS meas. value [Hg] ng‚g-1 SRM 2976 Mussel Tissue a

60.07

32.41 29.33 29.34 33.84 31.23 2.264 7.25

cert. value [Hg] ng‚g-1 61.0 ( 3.6

Meas., measured value; cert., certified value.

results do, however, reflect a marked improvement in data accuracy over implementing cold vapor generation ICPMS without isotope dilution, or traditional (solution) ICPMS, which in some cases can produce mercury values that differ from CVAAS values by as much as a factor of 2. These observations are confirmed by referring to a study conducted by Knight and co-workers, who used a nearly identical vapor-generation ICPMS experimental setup (but without isotope dilution) to analyze mercury samples on the order of 100-600 ng‚g-1 in the solid24 in human hair. This study showed that on average, ICPMS-derived mercury values were 18% lower than their corresponding CVAAS values. Further, Knight et al. showed that mercury concentration data determined by solution-based ICPMS was >2× lower than corresponding CVAAS data for CRM GBW 07601, a human hair standard. These observations, coupled with the fact that consistent mercury values were obtained for SRM 966 using both CVAAS and ID-CV-ICPMS, suggest that the ID-CV-ICPMS technique is a more robust method for ICP determination of mercury. This approach results in reduced memory effects and improved mercury signal-tobackground ratios (by at least a factor of 100) relative to previous in-house experience with ID-ICPMS solution analyses. The dry plasma conditions that were employed and the gaseous sample transport mechanism may partially explain the increased sensitivity and efficiency of the ID-CV-ICPMS technique. The concentration of mercury in the level I samples of SRM 966 was not accurately measured because of the extremely low natural level. Although adequate instrumental sensitivity was available, the concentration of mercury in the samples was less than that of the average blank (0.025 ng‚mL-1). The concentration of mercury in SRM 966 level I is