Determination of Mercury in Biological and Environmental Samples by Inductively Coupled Plasma Mass Spectrometry With the Isotope Dilution Technique JUN YOSHINAGA* AND MASATOSHI MORITA National Institute for Environmental Studies, 16-2 Onogawa, T sukuba, Ibaraki 305, Japan The accurate and precise determination of total mercury (Hg) the same CRMs to compensate for matrix effects arising from the high sodium content of the back-extracted sample.9 in biological and environmental samples by isotope dilutioninductively coupled plasma mass spectrometry (ID-ICP-MS) Campbell et al.10 determined total Hg in a cod muscle CRM by ID-ICP-MS for certification.They found good agreement is described. The precision of Hg isotope ratio (e.g., 200Hg/202Hg) measurements at the 20 ppb level was <0.5%. between the results obtained by ID-ICP-MS and those obtained by other methods including AFS. They also noted The deviation of the measured isotope ratios in a standard Hg solution was <0.5% from the natural ratios. Neither spectral that a stable isotope spike should be added to the sample before overnight digestion at room temperature to ensure interferences nor matrix effects were found to affect the accuracy and precision of the proposed Hg isotope ratio isotope equilibrium, which is followed by a pressurized microwave digestion.analysis. Total Hg concentrations in human hair and sediment CRMs were determined by ID-ICP-MS after 202Hg addition The National Institute for Environmental Studies (NIES), Japan Environment Agency, recently issued a human hair and acid decomposition of the samples.Solvent extraction and back-extraction was used for sediment analysis. The various reference material for Hg speciation.11 During the certification of the total Hg content of this CRM, ID-ICP-MS was employed decomposition methods used for hair samples, i.e., microwave digestion and the Teflon vessel double digestion method, did in this laboratory. The technique was also extended to the determination of total Hg in candidate sediment CRMs which not give any difference in the analytical value.The ID-ICP-MS results were better than those obtained with were prepared at the NIES. This paper describes the determination of Hg in human hair and sediment CRMs by standard additions or internal standardization in terms of accuracy and precision. Analytical results for human hair and ID-ICP-MS, which was considered to be the most accurate and precise analytical method for Hg.sediment CRMs were in good agreement with the certified/ reference values. Keywords: Isotope dilution-inductively coupled plasma mass EXPERIMENTAL spectrometry ; isotope ratio; total mercury; human hair; Instrumentation sediment The ICP-MS instrument used was a Hewlett-Packard (Avondale, PA, USA) HP-4500. Typical operating conditions Mercury (Hg) is commonly determined in environmental and biological samples by AAS or AFS after elemental Hg gener- are summarized in Table 1.The monitored masses were 200 ation and aeration. These analytical techniques are suitable and 202 for sample analyses and other masses (198, 199, 201 for routine analysis because of their high sensitivity and and 204) were also monitored in preliminary experiments on selectivity, reasonable accuracy and precision, and low run- the accuracy and precision of isotope ratio measurements with ning cost. this instrument.The ion lens voltage settings and other param- ICP-MS may be the method of choice for Hg determination eters of the instrument were tuned daily to obtain a 205Tl in certain circumstances, such as in the certification of reference signal of about 100 000 counts s-1 (10 ppb). materials where analytical values from several analytical A pyrolysis-Au amalgamation-AAS (PAAS) system methods with different analytical principles are required. The (MA1-S/MD-1, Nippon Instruments, Tokyo, Japan) was also capability of isotope analysis by ICP-MS offers further suit- used for Hg determination.ability because it permits accurate and precise determination of Hg by the stable isotope dilution (ID) technique. Indeed, ID-ICP-MS has frequently been employed in the certification Table 1 Operating conditions for ICP-MS of element contents in CRMs.1–5 Although ID-ICP-MS has been applied to the determination of a number of elements in Instrument HP-4500 (Hewlett-Packard) Rf power 1.2 kW various matrices, only a few applications have been reported Reflected power <5 W for the determination of Hg.This is surprising because ICP is Plasma gas flow rate 18.0 l min-1 a superior ionization source even for elements with high Nebulizer gas flow rate 1.10 l min-1 ionization potentials (Hg is a typical element), and the isotope Detector voltage -2050 V analysis of Hg, including the ID technique, is an important Sampling cone 0.8 mm diameter (nickel) area of application of ICP-MS.Skimmer cone 0.5 mm diameter (nickel) Sample uptake rate 0.2 ml min-1 (by peristaltic pump) Beauchemin and co-workers6–8 reported the use of multi- Spray chamber Scott type (Pyrex) element ID-ICP-MS, which included Hg as one of the analytes, Nebulizer Concentric type (Pyrex) in the analysis of marine sediment and biological CRMs. The Data acquisition 3 points per mass, 3 s per mass same group also used the ID technique combined with flow Monitored m/z 198, 199, 200, 201, 202, 204 injection for the ICP-MS determination of methylmercury in Journal of Analytical Atomic Spectrometry, April 1997, Vol. 12 (417–420) 417Reagents One gram of sediment sample, along with an appropriate amount of stable isotope spike, was placed in a 50 ml glass A stock standard Hg solution (1000 mg g-1) was prepared by centrifuge tube and decomposed with nitric acid (4 ml) for dissolving high purity (99.999%) mercury chloride (Soekawa 30 min in a water-bath at 95 °C.Hydrochloric acid (5 ml) was Kagaku, Tokyo, Japan) in 0.3 M nitric acid. Working standard then added and heating was continued for a further 20 min. solutions were prepared daily from the stock solution. The After cooling, 10 ml of a 50% solution of diammonium hydro- stock solution was stored in a Teflon bottle at 4 °C in the dark. gen citrate were added to prevent hydroxide formationfollowed The nitric acid used for sample digestion was of ‘Ultrapure’ by the addition of 10 ml of concentrated ammonia solution.grade (Kanto Chemical, Tokyo, Japan). Hydrochloric acid, Then, 1 ml of a 20% aqueous potassium iodide solution was ammonia solution and 4-methylpentan-2-one (IBMK) were of added. At this point the pH of the solution was 9.5–10.5 as atomic absorption grade from the same manufacturer. High measured by indicator paper. The Hg in the aqueous layer purity potassium iodide (99.9%), atomic absorption grade was extracted into 10 ml of IBMK by mechanically shaking diammonium hydrogen citrate and reagent-grade cysteine were the capped centrifuge tube for 30 min.A portion (2–4 ml) of purchased from Wako Chemical (Osaka, Japan) and were used the IBMK layer was withdrawn into a 10 ml glass test-tube in the solvent extraction of Hg from the acid digest of the fitted with a screw cap and subjected to back-extraction with sediment CRM sample. 2–4 ml of 20 mM aqueous cysteine solution, prepared daily, The 202Hg spike was prepared by dissolving 202HgO powder using a vortex mixer for 1 min.The IBMK layer was removed (Oak Ridge National Laboratory, Oak Ridge, TN, USA) in with a Pasteur pipette and the cysteine layer was analysed 1 M nitric acid. The spike solution was stored in a Pyrex glass by ICP-MS. bottle. The accurate 202Hg concentration of the spike solution was determined by a reverse ID technique against an Hg standard solution, prepared as described above, and was found Measurement to be 184 mg g-1.Repeated reverse ID analyses indicated that the concentration did not change during the study period In ID-ICP-MS measurements, the isotope ratio (200Hg/202Hg) (>1 year). of the spiked sample was measured under the conditions shown A 1M hydrobromic acid solution was prepared from a high in Table 1. A blank and a 20 or 50 ppb standard solution of purity reagent (AA-1,000, Tama Chemical, Kawasaki, Japan) natural abundance were measured prior to and after every 3–5 and used for the washing of the sample introduction system samples.The mass discrimination factor was changed when a of the ICP-MS instrument. Millipore purified water was used significant deviation (>0.5%) of the isotope ratio was found for the dilution of standards and decomposed samples. in the standard solution. Such deviation occurred once or twice a day, and was probably due to instrumental temperature- dependent drift of the mass calibration.The altered isotope Samples ratios obtained by the addition of the stable isotope spike were Human hair CRMs from the NIES (NIES CRM No. 5 and 13), BCR (CRM 397) and Shanghai Institute of Nuclear used for the calculation of the Hg concentration in the sample Research (GBW 09101) were used. The sediment CRMs ana- according to the conventional equation.1–10 lysed were NIES CRM No. 2 Pond Sediment, NRCC MESS-1 In the SA method, only 202Hg was monitored using a 3 s and BCSS-1, and NIST (formerly NBS) SRM 1645 River integration of the ion counts.In the IS method, in addition to Sediment. The moisture content of these reference materials 202Hg (3 s integration), 195Pt and 205Tl were also monitored was determined as specified by each organization. NIES candi- using a 1.5 s integration. date CRM No. 12 Marine Sediment for organotin compounds and No. 16 River Sediment for polyaromatic hydrocarbons RESULTS AND DISCUSSION were also analysed.Memory Effect Procedure It is known that the memory effect of Hg increases the blank The hair sample, typically 50 mg, was decomposed by the Teflon vessel double digestion method,12 after adding an counts and worsens the analytical performance of ICP-MS. To reduce the blank counts due to the memory effect, prolonged appropriate amount of the 202Hg spike solution (volume: 100 ml), with 1–1.5 ml of nitric acid. This digestion method is washing of the sample introduction system is often necessary.For example, when a 20 or 50 ppb standard solution was suitable for the determination of volatile elements because the digestion is completed in a doubly closed system. The digest introduced for 3 min, the blank count increased to 10% of the count of the standard solution for a prolonged period when was diluted to 10 g with water in a Pyrex glass test-tube with a screw cap. Microwave digestion of hair samples was also dilute nitric acid was used for washing.It was also found that a pure standard solution produced a greater memory effect tested. The 202Hg spike was added to 50 mg of NIES CRM No. 13 and samples were heated in a microwave oven either than a sample solution of the same Hg concentration. The memory effect severely increased the detection limit from 0.02 immediately afterwards or after overnight digestion at room temperature. The digestion procedure was essentially that ppb under ‘clean’ conditions to 0.2 ppb (3s definition at m/z= 202).It also adversely affected the accuracy and precision of described by Isoyama et al.13 using 1.5 ml of nitric acid. The microwave oven used was for domestic use; the pressurized Hg isotope ratio measurements unless the blank count was measured prior to each sample and standard. digestion vessel, consisting of a Teflon inner vessel (25 ml) and poly(propylene) jacket, was a San-ai Kagaku P-25 (Nagoya, It has been suggested that dilute hydrobromic acid is effective in reducing the memory effect of Hg.15 According to our Japan).A 50 mg amount of unspiked NIES CRM No. 13 was experience, 0.5–1 M hydrobromic acid reduced the blank count more efficiently than nitric, hydrochloric or sulfuric acid or digested by the Teflon vessel double digestion method for Hg determination by ICP-MS with standard additions (SA) and 20 mM cysteine solution. Introduction of 1 M hydrobromic acid for 3 min at high peristaltic pump speed (0.5 rev min-1 or internal standardization (IS) methods.After decomposition and dilution, Hg standard solution was added at levels of 20 0.7 ml min-1) decreased Hg blank counts to a tolerable level (1% of the standard solution) after introduction of a high Hg and 40 ng g-1 or internal standard (Pt and Tl) mixture at a level of 10 ng g-1. standard solution (e.g., 50 ppb) for 3 min. Therefore, 1 M hydrobromic acid washing was employed after analysis of each Sample preparation for sediment CRMs was based on the method of Sanzolone and Chao14 with some modifications.sample and standard. 418 Journal of Analytical Atomic Spectrometry, April 1997, Vol. 12Table 2 Accuracy and precision of Hg isotope ratio measurements by ICP-MS 198/202 199/202 200/202 201/202 204/202 Day 1 0.334 (0.6)* 0.5655 (0.1) 0.7754 (0.3) 0.4408 (0.2) 0.2290 (0.2) Day 2 0.335 (0.7) 0.5649 (0.4) 0.7746 (0.1) 0.4422 (0.4) NA† Day 3 0.332 (0.8) 0.5689 (0.6) 0.7741 (0.4) 0.4447 (1.2) NA Day 4 NA 0.5674 (0.2) 0.7749 (0.2) 0.4418 (0.3) NA Mean (RSD)‡ 0.334 (0.5) 0.5667 (0.3) 0.7748 (0.1) 0.4424 (0.4) — Theoretical§ 0.334 0.5650 0.7736 0.4414 0.2301 * Mean (RSD, %) of repetitive measurements (n=3–5) in one analysis day.† Not analysed. ‡ Mean (RSD, %) of means of Day 1 to Day 4. § From IUPAC.16 Table 3 Comparison of Hg concentration (mg g-1 dry mass) in NIES CRM No. 13 human hair determined by ID-ICP-MS, SA-ICP-MS and IS-ICP-MS.* n=5 in all instances ID-ICP-MS SA-ICP-MS IS-ICP-MS (Pt) IS-ICP-MS (Tl) Certified value 4.30±0.05 4.25±0.43 4.25±0.03 4.05±0.04 4.42±0.20 * SA-ICP-MS: ICP-MS determination of Hg with standard additions method.IS-ICP-MS: ICP-MS determination of Hg with internal standardization. Accuracy and Precision of Isotope Ratio Analysis of Hg by precision compared with the ID and IS techniques because of instrumental sensitivity drift during the analysis. Moreover, ICP-MS the number of samples required for the SA method is 2–3 Crude values of measuredHg isotope ratios were not consistent times the number required for the ID or IS technique, leading with natural abundance ratios because of the mass discrimi- to poor sample throughput.From this result, it was concluded nation of the instrument. Therefore, mass discrimination was that ID-ICP-MS was the best ICP-MS quantification method corrected daily by the periodic analysis of a Hg standard in terms of precision, accuracy and sample throughput. solution (20 or 50 ppb).The Hg isotope ratios reported here- Table 4 shows a comparison of the Hg concentration in a after are mass discrimination-corrected values. human hair CRM (NIES CRM No. 13) determined by The accuracy and precision of Hg isotope ratio analysis by ID-ICP-MS after three different sample digestion procedures, ICP-MS was examined using a Hg standard solution (20 ppb i.e., microwave digestion after overnight standing, microwave in 0.1 M nitric acid) with natural isotopic composition.The digestion immediately after spike addition, and the Teflon results are shown in Table 2. At this concentration, the 202Hg vessel double digestion method. This comparison was made count was typically about 20000 counts s-1 (total count because Campbell et al.10 reported that, for the ID-ICP-MS >60 000 per 3 s accumulation). Isotope ratios were measured determination of Hg in fish tissue, the sample should be 3–5 times a day for 3–4 d. Thus, the within- and between-day digested overnight at room temperature after the addition of variation in the measured isotope ratios could be examined.the stable isotope spike and nitric acid to the sample before The precision was generally <1% for both within- and microwave heating. They stated that accurate results could not between-day measurements. The deviations from the ‘true’ be obtained if the spike was added to the sample immediately values16 were all<0.5%. From these results it can be concluded before microwave heating, because isotope equilibrium could that ICP-MS measurements of Hg isotope ratios are sufficiently not be established between organically bound Hg in the sample precise and accurate for ID analysis.and the inorganic Hg spike. However, as is evident from In the m/z region of Hg no spectral interferences originating Table 4, the three digestion procedures did not produce any from molecular ions are expected. This was confirmed by the differences in the analytical values for Hg in the hair CRM in observation that the isotope ratios of Hg in digested hair were the present study.It is possible that the present microwave consistent with those in a standard solution. The only exception digestion procedure is more efficient for the destruction of the was 204Hg/202Hg, where an isobaric interference from 204Pb organic matrix and liberation of inorganic Hg from the sample exists. This result indicates that unpredictable spectral inter- matrix than the procedure of Campbell et al.The difference in ferences or other influences from the matrix are negligible in sample matrix (fish tissue in the work of Campbell et al. and isotope ratio measurements of Hg (except for those involving hair in the present study) may also be a cause of inconsistent 204Hg) by ICP-MS. results although methylmercury is the dominant Hg species in both types of sample. Table 5 presents ID-ICP-MS results for human hair CRMs, Determination of Total Hg in Biological CRMs using the Teflon vessel double digestion method.This digestion Table 3 compares the analytical results for NIES CRM No. 13 procedure was chosen because it is an established method and obtained by ID-ICP-MS, SA-ICP-MS and IS-ICP-MS (Tl and Pt). As can be seen, the mean values obtained by ID-ICP-MS, SA-ICP-MS and Pt-IS-ICP-MS were all similar and within Table 4 Comparison of Hg concentration (mg g-1 dry mass) in human the range of uncertainty of the certified value.Internal stan- hair CRM determined by ID-ICP-MS after three different digestion procedures dardization based on Tl gave lower values than the other methods. This might be due to the difference in the extent of Microwave-1* Microwave-2† Electric oven‡ the matrix effect between Hg and Tl. This is probable since 4.31±0.05 (n=3) 4.31±0.14 (n=4) 4.31±0.07 (n=4) the ionization potential of Hg (10.4 eV) is much higher than that of Tl (6.1 eV) and this is an inherent problem of the IS * Microwave-1: Pressurized digestion by microwave irradiation after technique. Although Pt was found to be a better internal addition of stable isotope spike and acid and overnight standing at standard in the present study, probably because its ionization room temperature.† Microwave-2: Pressurized digestion by micro- potential (9.0 eV) is fairly similar to that of Hg, this does not wave irradiation immediately after the addition of stable isotope spike mean that Pt will be the best internal standard for other and acid.‡ Electric oven: Pressurized digestion at 140 °C for 4 h in an electric oven. sample matrices. The present SA-ICP-MS method gave poor Journal of Analytical Atomic Spectrometry, April 1997, Vol. 12 419Table 5 Mercury concentrations in human hair CRMs determined Table 6 Mercury concentrations in sediment CRMs by ID-ICP-MS (mg g-1 dry mass) by ID-ICP-MS (mg g-1 dry mass)* Found† Certified Certified/reference Found* value NIES CRM No. 5 4.40±0.08 (n=6) 4.4±0.4 NIES CRM No. 13 4.30±0.06 (n=15) 4.42±0.20 0.987±0.031 (3) 1.1±0.5 NIST SRM 1645 River Sediment BCR CRM 397 11.8±0.2 (n=6) 12.3±0.5 GBW 09101 1.96±0.05 (n=3) 2.16±0.21 NRCC MESS-1 0.174±0.004 (3) 0.171±0.014 NRCC BCSS-1 0.132±0.001 (3) 0.129±0.012 NIES CRM No. 2 1.22±0.02 (3) (1.3)† * Samples (50 mg) were digested with 1–1.5 ml of nitric acid by the Teflon vessel double digestion method at 140 °C for 4 h after addition Pond Sediment NIES candidate CRM 1.10±0.02 (6) 1.22‡ of stable isotope spike.† Mean±standard deviation of repetitive analyses. Number of analyses is indicated in parentheses. No. 12 Marine Sediment NIES candidate CRM 0.866±0.008 (6) 0.95‡ is routinely employed in our laboratory, even though it may No. 16 River Sediment be more time consuming. The ID-ICP-MS results were in * Mean±standard deviation of repetitive analyses. Number of analy- good agreement with the certified values of the CRMs, with ses is indicated in parentheses. † Reference value.‡ Solid sampling- relative standard deviations of 1–2%. This result further pyrolysis-Au amalgamation-AAS. Aqueous Hg standard was used for demonstrates that ID-ICP-MS analysis offers an accurate and calibration. precise determination of Hg in hair samples. back-extraction process. However, the reason for the poor Determination of Total Hg in Sediment CRMs recovery of Hg from NIST SRM 1645 was not pursued further because the accuracy of ID analysis is independent of the The sample decomposition method for geological samples recovery rate if isotope equilibrium is established.reported by Sanzolone and Chao14 was used for sediment analysis. Two major modifications were made to the original method; firstly, a 50% solution of diammonium hydrogen CONCLUSION citrate was added prior to adjusting the sample digest to ID-ICP-MS offers accurate and precise determination of Hg alkaline pH.Without this addition, a dense hydroxide layer in biological and sediment samples. It is more accurate and formed and caused problems in withdrawing an aliquot of the precise than other modes of determination by ICP-MS. The IBMK layer for back-extraction. Secondly, since it is not application of MS to Hg determination has been limited practical to introduce IBMK into an ICP-MS system, back- because the ionization of Hg by conventional ionization extraction from the IBMK layer was employed. The Hg–I sources is not efficient.Therefore, the capability of the isotope complex was very stable in IBMK and quantitative back- analysis of Hg is a unique feature of ICP-MS. This technique extraction was achieved only with 20 mM cysteine solution and is of great value as an alternative method to AAS or AFS not with 1 M nitric or hydrobromic acid. A standard solution particularly in the certification of environmental and biologi- of Hg was pre-treated in a similar way to the sample and the cal CRMs.recovery was examined; the average recovery was 105%. However, since the solubility of IBMK in the aqueous layer is The authors thank K. Takata and C. Komatsu, NIES, for the fairly high, up to 70% suppression in the Hg ion counts from operation of the ICP-MS and PAAS systems. the dissolved IBMK was found after the introduction of several back-extracted samples. Therefore, SA or IS is necessary for REFERENCES the determination of Hg in back-extracted samples if the determination is based on external calibration.The use of a 1 McLaren, J. W., Beauchemin, D., and Berman, S. S., Anal. Chem., 1987, 59, 610. less water-soluble extraction solvent may be recommended in 2 Beauchemin, D., McLaren, J. W., Mykytius, A. P., and Berman, this case. However, the suppression does not affect the accuracy S. S., Anal. Chem., 1987, 59, 778. of Hg isotope ratio measurements because it affects all of the 3 Okamoto, K., Sci.T otal Environ., 1991, 107, 29. Hg isotopes to the same extent and this is certainly the 4 Okamoto, K., Spectrochim. Acta, Part B, 1991, 46, 1615. advantage of the ID technique. 5 Murphy, K. E., and Paulsen, P. J., Fresenius’ J. Anal. Chem., Table 6 shows analytical results for sediment CRMs by 1995, 352, 203. 6 Beauchemin, D., McLaren, J. W., Willie, S. N., and Berman, S. S., ID-ICP-MS. The measured ratio was 200Hg/202Hg. The agree- Anal. Chem., 1988, 60, 687.ment between the found values in NIST SRM 1645 River 7 McLaren, J. W., Beauchemin, D., and Berman, S. S., Spectrochim. Sediment, NRCC MESS-1 andNRCC BCSS-1 and the respect- Acta, Part B, 1988, 43, 413. ive certified values was satisfactory. However, a small deviation 8 Beauchemin, D., McLaren, J. W., and Berman, S. S., J. Anal. At. was found for the analytical values of the NIES candidate Spectrom., 1988, 3, 775. sediment CRMs obtained by ID-ICP-MS from those derived 9 Beauchemin, D., Siu, K. W. M., and Berman, S. S., Anal. Chem., 1988, 60, 2587. from PAAS. The reason for this is not clear. Cooperative 10 Campbell, M. J., Vermeir, G., Dams, R., and Quevauviller, Ph., analytical data on these CRMs are not yet available. J. Anal. At. Spectrom., 1992, 7, 617. As mentioned above, the recovery of Hg from standard 11 Yoshinaga, J., Morita, M., and Okamoto, K., Fresenius’ J. Anal. solutions was close to 100%. However, the recovery was Chem., 1996, in the press. poorer (around 50%) for NIST SRM 1645 River Sediment as 12 Okamoto, K., and Fuwa, K., Anal. Chem., 1984, 56, 1758. estimated from the 200Hg count. PAAS analysis showed that 13 Isoyama, H., Uchida, T., Oguchi, K., Iida, C., and Nakagawa, G., Anal. Sci., 1990, 6, 385. all of the Hg present in this SRM was liberated by nitric acid 14 Sanzolone, R. F., and Chao, T. T., Analyst, 1983, 108, 58. decomposition, employed as described under Experimental, 15 Beary, E. S., NIST, 1996, personal communication. even without the addition of hydrochloric acid. As the 202Hg 16 IUPAC, Pure Appl. Chem., 1991, 63, 991. spike and all of the intrinsic Hg were treated with a mixture of hot nitric and hydrochloric acid for at least 20 min during Paper 6/06171K the decomposition procedure, it is highly probable that during Received September 9, 1996 this process isotope equilibrium was established. Therefore, Accepted November 19, 1996 the poor recovery probably resulted from the extraction or 420 Journal of Analytical Atomic Spectrometry, April 1997, Vol. 12