Anal. Chem. 2004, 76, 7131-7136
Determination of Methylmercury in Environmental Matrixes by On-Line Flow Injection and Atomic Fluorescence Spectrometry Chun-Mao Tseng,*,† Chad R. Hammerschmidt,‡ and William F. Fitzgerald‡
National Center for Ocean Research, National Taiwan University, P.O. Box 23-13, Taipei 106, Taiwan, and Department of Marine Sciences, University of Connecticut, Groton, Connecticut 06340
The precision and bias of monomethylmercury (MMHg) determinations in environmental samples can be improved by directly coupling and automating the numerous steps involved with analysis of this toxic Hg species. We developed a simple and robust mercury speciation analyzer (MSA) for measurement of MMHg in environmental matrixes. This on-line hyphenated system couples the main analytical steps, including sample introduction, aqueous-phase ethylation, Tenax preconcentration, and gas chromatographic separation, to cold vapor atomic fluorescence detection and data acquisition. Here we describe the MMHg-MSA, present results of laboratory optimization and performance tests, and compare the reproducibility between dual analytical channels. With alternating sample concentration and analysis, a dualchannel system permits six high-accuracy MMHg determinations per hour. Additional advantages compared to the traditional manual method include ease of operation and high precision (18 MΩ cm-1) and aqueous sample are added to a borosilicate glass GLS so that the total volume of liquid in the GLS is 50-150 mL; no additional water is typically added for the analysis of MMHg in natural waters. The solution pH is adjusted to ∼4.9 with 2 M acetate buffer, and 200-600 µL of 1% (w/v) NaTEB (NaB(CH2CH3)4; Strem Chemical, Newburyport, MA) is added as the ethylating agent (Table 1).10 The NaTEB reacts with MMHg and (29) Tseng, C. M.; Balcom, P.; Lamborg, C.; Fitzgerald, W. F. Environ. Sci. Technol. 2003, 37, 1183-1188.
Figure 2. Recommended operation schedule of two channels for Hg speciation analysis with the MMHg-MSA.
Hg(II) in solution to produce ethylmethylHg (the MMHg derivative) and diethylHg (the Hg(II) derivative), which are volatile and can be stripped from solution. Promptly after addition of sample and reagents, the upper stopper of the GLS (Figure 1) is emplaced and the ethylation reaction is allowed to proceed for ∼2 min (Figure 2) while a flow of N2 carrier/stripping gas (130 mL min-1) is maintained through the headspace. N2 carrier/stripping gas is cleansed of Hg species before it enters the GLS by passage through both Au-coated sand and Carbotrap. After the reaction period, the N2 carrier/stripping gas is routed through a glass frit (∼20 µm porosity) at the bottom of GLS via the control of a threeway valve on the GLS (Figure 1). In this flow position, the N2 purges the volatile, ethylated Hg derivatives from solution. The gas flowing out of the GLS first passes through a precleaning trap (quartz or Teflon tubing, 3.2-mm i.d., 10-cm length) packed with reagent-grade soda lime (14-24 mesh, Fisher Scientific). The soda lime neutralizes any potential acidity and helps remove water vapor from the stripping/carrier gas. Ethylated Hg species in the gas stream are trapped and concentrated on Tenax (23% graphitized carbon, 20/35 mesh, Alltech) in a quartz tube (13-cm length, 3.2mm i.d.) while the stripping/carrier gas is vented out of the system and to a fume hood. Complete purging of ethylated Hg species from solution takes 8 min at a N2 flow of 130 mL min-1 for a 50150-mL sample (described below). After an 8-min purging period, the flow of N2 gas is returned to the headspace of GLS by turning the three-way valve. The flow of N2 through the headspace is maintained for 2 min to help remove water vapor that may have accumulated on the Tenax trap during collection of sample MMHg (Table 1, Figure 2). The 10-way valve is rotated from the collection position to the analytical position after this drying step, and the water is drained from the GLS through the 2-way valve (Figure 1). Thermal Desorption, Separation, and Detection. Derivatized Hg species collected on the Tenax column are analyzed by CVAFS. After concentration of ethylated Hg derivatives on the Tenax trap, the 10-way valve is alternated so that (1) the N2 carrier/stripping gas bypasses the Tenax trap and (2) Ar carrier gas (90 mL min-1, regulated by CVAFS unit) passes over the Tenax to an in-line isothermal GC column (U-shaped borosilicate glass or Teflon tube; 30-cm length, 3.2 mm) packed with Chromosorb WAW-DMSC (60/80 mesh) that is coated with 15% OV-3 (Supelco). The GC column is housed in a thermally insulated oven maintained at 50-80 °C (Figure 1, Table 1). The direction of the Ar flow through the Tenax trap is opposite to the previous flow direction of N2 carrier/stripping gas; this avoids passing desorbed Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
7133
Hg species down the entire length of the heated Tenax trap. Ar gas is passed through the Tenax column for 2 min to permit gas pressure stabilization and additional drying of the Tenax trap before thermal desorption and analysis of Hg species. Derivatized Hg species are desorbed from the Tenax trap by heating with a Nichrome resistance wire, which is wrapped around the column, for 2 min at 200-300 °C under continuous Ar flow. The Ar carries desorbed Hg derivatives to the isothermal GC column. The ethylderivatized Hg species are eluted sequentially from the GC column, on the basis of their molecular weights, and decomposed to Hg0 in a pyrolysis column upon elution. The pyrolysis column is a quartz tube, packed with quartz wool, that is heated at 600800 °C with Nichrome resistance wire inside a thermally insulated oven. Hg0 produced by the pyrolysis column is carried to the CVAFS detector, and the intensity of its fluorescence at 253.7 nm is quantified. The peak area of each Hg species is recorded and analyzed with either an integrator (e.g., Hewlett-Packard 3396A) or a personal computer with integration software (e.g., LabVIEW). A fan (not shown in Figure 1) is used to cool the Tenax column for 1 min following the 2-min period of thermal desorption. The entire analytical process, from flow stabilization to data acquisition, requires 10 min (Figure 2), and it is controlled by a programmable timer/controller (e.g., ChronTrol). Combined with the purging and preconcentration steps, the analysis of one sample requires 20 min. The use of dual analytical channels (i.e., two individual GLS systems coupled with Tenax traps) optimizes analytical efficiency by allowing alternating purging and analytical steps. That is, while derivatized Hg species are purged and concentrated in one analytical channel, the other channel is positioned for GC separation and analysis. Thus, sample throughput can be as high as six samples per hour when two analytical channels are operated simultaneously. An operation schedule for optimal analytical efficiency with two channels is shown in Figure 2. Standards and Samples. A standard solution of 1000 µg mL-1 Hg(II), traceable to the U.S. National Institute of Standards and Technology (NIST) was purchased, and a 1000 µg mL-1 solution of MMHg was prepared by dissolving methylmercuric chloride salt (Alfa Chemical) in propanol. Both solutions were stored in a refrigerator and protected from light. Working standard solutions of Hg(II) and MMHg (1 ng mL-1 in 1% HNO3) were prepared regularly by quantitatively diluting the concentrated standards. Working standards of MMHg were calibrated against Hg0 standards taken from the headspace over pure liquid after oxidation with BrCl.30 MMHg was determined in multiple environmental matrixes to validate the proposed methodology. Surface waters were sampled from small ponds in rural areas near Taipei, Taiwan, in March 2003. Rainwater (50-150 mL/sample) was collected near the National Center for Ocean Research in Taipei on April 1-10, 2003. Two steaks of fresh tuna were purchased from fish markets in Taipei. In addition, we analyzed MMHg in two certified reference materials, polluted estuarine sediment IAEA-356 (International Atomic Energy Agency) and lobster hepatopancreas TORT-2 (National Research Council of Canada). (30) Bloom, N. S.; Crecelius, E. A. Mar. Chem. 1983, 14, 49-59.
7134
Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
Blanks and Calibration. Blanks are critical to the accurate measurement of Hg species in environmental samples.31 Samples may be compromised by impurities in reagents and by contamination during collection, storage/preservation, and analysis. Impurities in NaTEB, for example, may cause the formation of artifact MMHg during the ethylation reaction.23,24 Therefore, reagents used in this study were purified prior to their use.6 Analytical blanks were defined as the amount of MMHg resulting from the purging of reagent-grade water containing the analytical reagents (i.e., acetate buffer and NaTEB). Analytical blanks were measured every five samples during this study, and they contained e2 pg of Hg as MMHg. All analytical glassware, tubing, and fittings were thoroughly cleaned with acid and rinsed with reagent-grade water before use. Details on proper preparation of soda lime traps, Tenax traps, and GC columns are described elsewhere.10,15,32,33 Newly prepared Tenax traps were heated to 300 °C for 10 min to remove any residual Hg prior to use; the useful longevity of the Tenax is dependent upon frequency of use and replacement frequency/ efficacy of the soda lime traps. Wetted soda lime traps can be dried by heating at ∼150 °C for 5 min prior to analyzing samples (recommended for soda lime in quartz tubing only). No silanization procedure is needed for the GC column prior to its use;23 however, it should be preconditioned with flowing Ar at ∼60 °C for at least 12 h prior to use. MMHg analyses were calibrated by adding aliquots of an aqueous working standard to reagent-grade water and analytical reagents in the GLS. Calibration curves were prepared for each GLS/Tenax channel, and the calibration was checked periodically during sample analysis. The MMHg-MSA yields excellent linear calibration regressions (r2 > 0.995), and the precision associated with this hyphenated approach exceeds that of the conventional methodology. The precision of replicate analyses of the same sample (analytical precision) was less than 5% relative standard deviation (RSD) for all samples analyzed during this study. This result was consistent regardless of whether the same or different analytical channels were used for replicate analyses. The accuracy of MMHg determinations made during this study also was assessed by routine analyses of (1) spiked subsamples and matrix blanks, (2) procedurally replicated subsamples from the same parent material, and (3) certified reference materials. Mercury Extractions. MMHg was extracted from environmental samples with commonly used methods. Aqueous distillation was used to extract MMHg from sediments and 0.45-µm filtered pond waters.23,24,27 Lyophilized biological tissues were digested with 25% tetramethylammonium hydroxide (TMAH)6,11 and diluted with reagent-grade water prior to MMHg analysis. For total Hg analysis, separate aliquots of biological tissue were digested with a 1:1 (v/v) mixture of HNO3 and H2SO4 with microwave irradiation.34 MMHg in rain samples was measured directly (i.e., no sample pretreatment). (31) Quevauviller, Ph.; Maier, E. A.; Griepink, B. In Quality Assurance for Environmental Analysis; Quevauviller, Ph., Maier, E. A., Griepink, B., Eds.; Elsevier: Amsterdam, 1995; pp 1-25. (32) Tseng, C. M.; de Diego, A.; Amouroux, D.; Donard, O. F. X., Chemosphere 1999, 39, 1119-1136. (33) Rodriguez Pereiro, I.; Wasik, A.; Lobinski, R. J. Anal. At. Spectrosc. 1998, 13, 743-747. (34) Uhrberg, R. Anal. Chem. 1982, 54, 1906-1908.
Mercury Determinations. MMHg was measured in rainwater, sediment and pond water distillates, and digestates of biological materials. Aliquots of the extracts were added to a GLS and diluted with reagent-grade water; samples of rainwater were neither extracted nor diluted. Samples were amended with 400 µL of 2 M acetate buffer and 200-600 µL of NaTEB solution. MMHg and ionic Hg (i.e., Hg(II)) in the samples were measured with the MSA, using the operating conditions discussed above and outlined in Table 1. RESULTS AND DISCUSSION Optimization of the MMHg-MSA. Careful design of an online hyphenated system can result in enhanced sensitivity, high reproducibility and accuracy, a low detection limit, and rapid analysis (i.e., six samples per hour). In this study, we focused on the optimization of physical variables of the MSA instead of wet chemical parameters (e.g., volume of derivatizing agent, reaction pH, desorption and pyrolytic atomization temperatures, and trapping efficiency), which have been studied by others.10,15 Our experiments focused on the following three areas: chromatographic separation of Hg species, purging efficiencies, and evaluation of general analytical performance (i.e., detection limit, precision, and bias). Chromatographic Separation of Hg Species. Hg species were derivatized with NaTEB.10,15 MMHg and Hg(II) reacts with NaTEB to yield the volatile Hg species methylethylHg (MeHgEt, the MMHg derivative) and diethylHg (HgEt2, the Hg(II) derivative). Both species are identified by their retention times after GC separation. Peak resolution is dependent mostly on the flow rate of carrier gas (Ar), the GC heating program, and the quality of GC column packing. Optimization experiments showed that good resolution of the two Hg species, with Gaussian peak symmetry and high sensitivity, can be achieved with an Ar flow rate of 80-100 mL min-1 and a GC column temperature of 5080 °C. These conditions result in no analyte decomposition. Lesser flow rates of Ar carrier gas result in a longer elution (analytical) time and expansion and tailing of chromatographic peaks. Greater GC column temperatures can result in coelution of the two derivatized Hg species or potential analyte degradation to Hg0. Figure 3 illustrates three typical chromatograms of derivatized Hg species obtained with the MMHg-MSA. Each chromatogram shows good baseline separation of MeHgEt and HgEt2 and excellent peak symmetry (i.e., absence of peak tailing). The broadening of the HgEt2 peak in each of these chromatograms, compared to MeHgEt, is due to the relatively lower volatility of this analyte, and it is a characteristic effect of isothermal GC.15 Temperature-gradient GC would improve the chromatograph of HgEt2; however, its use is impractical when sample through-put and high-accuracy determinations of MMHg (i.e., MeHgEt), specifically, are the primary objectives. MeHgEt elutes from the column at ∼2.6 min and HgEt2 at ∼6 min, with the described analytical conditions (Table 1). Slightly varied retention times may be obtained due to differences in GC conditions (e.g., carrier flow rate, column temperature, packing quality). An 8-min period of chromatographic separation was used because it corresponds to the same amount of time needed for sample purging and trapping with the other analytical channel. Purging Efficiencies. We examined the effects of purging time and stripping gas (N2) flow rate on the recovery of derivatized
Figure 3. Typical chromatograms of MeHgEt (MMHg derivative, peak 1) and HgEt2 (Hg(II) derivative, peak 2), obtained with the MSA under optimum analytical conditions (Table 1) from different matrixes: (a) synthetic water sample (85 pg of MMHg, 98 pg of Hg(II) in reagent-grade water), (b) TMAH digestate of lobster hepatopancreas (TORT-2), and (c) aqueous distillate of marine sediment (IAEA-356).
Figure 4. Effects of purging time (circles) and N2 flow rate (squares) on the purging efficiency for MMHg. Purging time tests were conducted with 100 mL of aqueous solution at a constant purge rate of 130 mL min-1. Tests of purging flow rate were done with 100 mL of solution and a constant purging time of 8 min. Error bars are (1 standard deviation of the mean and were less than the symbol size where they are not evident.
Hg species from solution (Figure 4). Purging efficiency was optimized by analysis of a synthetic aqueous solution (i.e., reagentgrade water containing 85 pg of MMHg and 98 pg of Hg(II)) under optimal ethylation conditions with NaTEB (Table 1). Figure 4 shows that ethylated MMHg is completely removed from a 100mL aqueous sample in 8 min at a fixed N2 flow of 130 mL min-1. A shorter purging duration results in nonquantitative stripping of the volatile MMHg derivative, and a longer purging period, though yielding complete recovery, can result in excessive water condensation on the Tenax trap, which can interfere with CVAFS determinations.28 Alternatively, for a constant purging time of 8 min, a N2 flow rate of g130 mL min-1 is needed for complete recovery of MMHg from solution (Figure 4). Quantitative recovery of Hg(II) from solution also was obtained with the same purging rate and time (130 mL min-1, 8-min purge; data not shown). This Analytical Chemistry, Vol. 76, No. 23, December 1, 2004
7135
Table 2. Results Obtained during the Validation of the MMHg-MSA measured (µg g-1 dry weight) sample
MMHg
total Hga
Hg(II)
certified (µg g-1 dry weight) MMHg
total Hg
TORT-2 0.15 ( 0.02 0.13 ( 0.02 0.28 ( 0.04 0.152 ( 0.013 0.27 ( 0.06 Tuna-1 0.47 ( 0.02 0.02 ( 0.01 0.48 ( 0.05 Tuna-2 0.62 ( 0.04 0.02 ( 0.01 0.64 ( 0.06 (ng g-1 dry weight) IAEA-356
5.6 ( 0.5
b
(ng g-1 dry weight) 5.64 ( 0.038
a Total Hg determined by SnCl reduction after BrCl oxidation. b Not 2 detected or quantified.
Figure 5. Comparison of sample MMHg concentrations measured with both channel 1 and channel 2 (1:1 line plotted for reference). Error bars are (1 standard deviation of the mean and were less than the symbol size where they are not evident.
indicates similar kinetics for the ethylation reaction and gas-phase transfer of the two Hg species. These results are comparable to those found with the cryogenic-GC separation technique.6 Hence, ethylated derivatives of both MMHg and Hg(II) species are fully recovered from an aqueous solution with an N2 purging rate of 130 mL min-1 for an 8-min period. Evaluation of Analytical Performance. The MMHg-MSA was evaluated for recovery of MMHg in analytical blanks, detection limit, degree of reproducibility, and recovery of known additions. The trapping performance of each analytical channel in the dual-channel system was similar. Figure 5 shows that the recovery of MMHg is the same between analytical channels, regardless of sample type, extraction methodology, or analyte concentration. Excellent results from analyses of quality control samples with the MMHg-MSA validate this method. Low concentrations of MMHg were measured in analytical blanks. Determined amounts of MMHg and Hg(II) in analytical blanks were less than or equal to 2 and 10 pg, respectively. The absolute detection limit of the MMHg-MSA, defined as three times the standard deviation of the procedural blank, was about 1 pg for MMHg and 10 pg for Hg(II), during a method validation period of 10 analytical batches. Method detection limits are estimated at ∼0.01 ng g-1 for MMHg in a 0.1-g sample of dry sediment or biological tissue and ∼0.01 ng L-1 in aqueous samples (0.15 L). The precision of MMHg measurements with the MMHg-MSA averaged
